增加了HSMO整线优化方法，读取数据增加了供料器部分

.gitignore

@@ -9,3 +9,4 @@ Scripts/
 *.pkl
 *.pth
 *.m
+opt/

base_optimizer/optimizer_common.py

@@ -22,6 +22,7 @@ import pandas as pd
 import matplotlib.pyplot as plt
 import matplotlib
 import traceback
+import openpyxl
 
 matplotlib.use('TkAgg')
 
@@ -67,7 +68,16 @@ t_fix_camera_check = 0.12  # 固定相机检测时间
 T_pp, T_tr, T_nc, T_pl = 2, 5, 25, 0
 
 # 时间参数 （数据拟合获得）
-Fit_cy, Fit_nz, Fit_pu, Fit_pl, Fit_mv = 0.326, 0.870, 0.159, 0.041, 0.001
+# Fit_cy, Fit_nz, Fit_pu, Fit_pl, Fit_mv = 0.326, 0.870, 0.159, 0.041, 0.030
+Fit_cy, Fit_nz, Fit_pu, Fit_pl, Fit_mv = 0.326, 0.870, 0.159, 0.041, 0.035
+
+
+class Point:
+    def __init__(self, _x, _y, _r=0, _h=None):
+        self.x = _x
+        self.y = _y
+        self.r = _r
+        self.h = _h
 
 
 class OptResult:
@@ -108,9 +118,12 @@ class OptInfo:
         print(f'-Pick time: {self.pickup_time: .3f}, Pick distance: {self.pickup_distance: .3f}')
         print(f'-Place time: {self.place_time: .3f}, Place distance: {self.place_distance: .3f}')
         print(
-            f'-Round time: {self.total_time - self.place_time - self.place_time: .3f}, Round distance: '
+            f'-Round time: {self.total_time - self.operation_time - self.pickup_time - self.place_time: .3f}, Round distance: '
             f'{self.total_distance - self.pickup_distance - self.place_distance: .3f}')
 
+        print(f'-Round & place time per cycle: {(self.total_time - self.pickup_time - self.operation_time) * 1000.0 / (self.cycle_counter + 1e-10): .3f}, ', end='')
+        print(f'-Round & place distance per cycle: {(self.total_distance - self.pickup_distance) / (self.cycle_counter + 1e-10): .3f}')
+
         minutes, seconds = int(self.total_time // 60), int(self.total_time) % 60
         millisecond = int((self.total_time - minutes * 60 - seconds) * 60)
 
@@ -856,3 +869,10 @@ def random_division(num, div):
 
 def list_range(start, end=None):
     return list(range(start)) if end is None else list(range(start, end))
+
+
+def kth_indices_partition(num, kth):
+    if len(num) > kth:
+        return np.argpartition(num, kth)
+    else:
+        return np.array(range(len(num)))

base_optimizer/optimizer_interface.py

@@ -15,26 +15,22 @@ def base_optimizer(machine_index, pcb_data, component_data, feeder_data, params,
 
     if params.machine_optimizer == 'cell-division':  # 基于元胞分裂的遗传算法
         component_result, cycle_result, feeder_slot_result = optimizer_celldivision(pcb_data, component_data)
-        placement_result, head_sequence = greedy_placement_route_generation(component_data, pcb_data, component_result,
-                                                                            cycle_result, feeder_slot_result)
+        placement_result, head_sequence = place_allocate_sequence_route_generation(component_data, pcb_data,
+                                                                                   component_result, cycle_result,
+                                                                                   feeder_slot_result)
     elif params.machine_optimizer == 'feeder-priority':  # 基于基座扫描的供料器优先算法
         component_result, cycle_result, feeder_slot_result = feeder_priority_assignment(component_data, pcb_data,
                                                                                         feeder_data)
-        placement_result, head_sequence = greedy_placement_route_generation(component_data, pcb_data, component_result,
-                                                                            cycle_result, feeder_slot_result)
-        # placement_result, head_sequence = beam_search_route_generation(component_data, pcb_data, component_result,
-        #                                                                cycle_result, feeder_slot_result)
-        # placement_result, head_sequence = scan_based_placement_route_generation(component_data, pcb_data,
-        #                                                                         component_result, cycle_result,
-        #                                                                         feeder_slot_result)
-
+        placement_result, head_sequence = scan_based_placement_route_generation(component_data, pcb_data,
+                                                                                   component_result, cycle_result,
+                                                                                   feeder_slot_result)
     elif params.machine_optimizer == 'hybrid-genetic':  # 基于拾取组的混合遗传算法
         component_result, cycle_result, feeder_slot_result, placement_result, head_sequence = optimizer_hybrid_genetic(
             pcb_data, component_data, hinter=hinter)
 
     elif params.machine_optimizer == 'aggregation':  # 基于batch-level的整数规划 + 启发式算法
         component_result, cycle_result, feeder_slot_result, placement_result, head_sequence = optimizer_aggregation(
-            component_data, pcb_data)
+            component_data, pcb_data, hinter=hinter)
     elif params.machine_optimizer == 'genetic-scanning':
         component_result, cycle_result, feeder_slot_result, placement_result, head_sequence = optimizer_genetic_scanning(
             component_data, pcb_data, hinter=hinter)
@@ -43,15 +39,16 @@ def base_optimizer(machine_index, pcb_data, component_data, feeder_data, params,
             component_data, pcb_data, hinter=hinter)
     elif params.machine_optimizer == "two-phase":
         component_result, feeder_slot_result, cycle_result = gurobi_optimizer(pcb_data, component_data, feeder_data,
-                                                                              initial=True, partition=True,
+                                                                              initial=True, partition=False,
                                                                               reduction=True, hinter=hinter)
 
-        placement_result, head_sequence = greedy_placement_route_generation(component_data, pcb_data, component_result,
-                                                                            cycle_result, feeder_slot_result)
+        placement_result, head_sequence = place_allocate_sequence_route_generation(component_data, pcb_data,
+                                                                                   component_result, cycle_result,
+                                                                                   feeder_slot_result)
     else:
         raise 'machine optimizer method ' + params.method + ' is not existed'
 
-    print('----- Placement machine ' + str(machine_index) + ' ----- ')
+    # print('----- Placement machine ' + str(machine_index) + ' ----- ')
     opt_res = OptResult(component_result, cycle_result, feeder_slot_result, placement_result, head_sequence)
     # 估算贴装用时
     info = placement_info_evaluation(component_data, pcb_data, opt_res, hinter=False)
@@ -67,4 +64,6 @@ def base_optimizer(machine_index, pcb_data, component_data, feeder_data, params,
             f'result/{params.filename[:-4]}-{params.line_optimizer}-M0{machine_index} {params.save_suffix}',
             component_data, pcb_data, opt_res)
 
+    # output_optimize_result(f'{params.filename[:-4]}', component_data, pcb_data, opt_res)
+
     return info

base_optimizer/result_analysis.py

@@ -355,15 +355,12 @@ def output_optimize_result(file_path, component_data, pcb_data, optimizer_result
     if 'desc' not in output_data.columns:
         column_index = int(np.where(output_data.columns.values.reshape(-1) == 'part')[0][0])
         output_data.insert(loc=column_index + 1, column='desc', value='')
-    file_dir = file_path[:file_path.rfind('/') + 1]
-    if not os.path.exists(file_dir):
-        os.makedirs(file_dir)
 
-    output_data.to_excel(file_path + '.xlsx', sheet_name='tb1', float_format='%.3f', na_rep='')
+    output_data.to_csv('result/' + file_path + '.txt', sep='\t', float_format='%.3f', header=False, index=False)
 
 
 def optimization_assign_result(component_data, pcb_data, optimizer_result, nozzle_hinter=False, component_hinter=False, 
-                               feeder_hinter=False):
+                               feeder_hinter=False, placement_hinter=False):
     if nozzle_hinter:
         columns = ['H{}'.format(i + 1) for i in range(max_head_index)] + ['cycle']
 
@@ -428,6 +425,29 @@ def optimization_assign_result(component_data, pcb_data, optimizer_result, nozzl
         print(feedr_assign)
         print('')
 
+    if placement_hinter:
+        columns = ['H{}'.format(i + 1) for i in range(max_head_index)] + ['cycle']
+
+        placement_assign = pd.DataFrame(columns=columns)
+        for cycle, _ in enumerate(optimizer_result.placement_assign):
+            placement_assign.loc[cycle, 'cycle'] = 1
+            for head in range(max_head_index):
+                point = optimizer_result.placement_assign[cycle][head]
+                if point != -1:
+                    placement_assign.loc[cycle, 'H{}'.format(head + 1)] = 'P{}'.format(point)
+                else:
+                    placement_assign.loc[cycle, 'H{}'.format(head + 1)] = ''
+
+        headseq_assign = pd.DataFrame(columns=columns)
+        for cycle, headseq in enumerate(optimizer_result.head_sequence):
+            headseq_assign.loc[cycle, 'cycle'] = 1
+            for head in range(len(headseq)):
+                headseq_assign.loc[cycle, 'H{}'.format(head + 1)] = 'H{}'.format(headseq[head])
+
+        print(placement_assign)
+        print(headseq_assign)
+        print('')
+
 
 def placement_info_evaluation(component_data, pcb_data, optimizer_result, hinter=False):
     # === 优化结果参数 ===
@@ -555,28 +575,43 @@ def placement_info_evaluation(component_data, pcb_data, optimizer_result, hinter
 
             # 贴装路径
             if optimizer_result.placement_assign and optimizer_result.head_sequence:
+                head_angle = [0 for _ in range(max_head_index)]
                 for head in optimizer_result.head_sequence[cycle]:
                     index = optimizer_result.placement_assign[cycle][head]
                     if index == -1:
                         continue
                     mount_pos.append([pcb_data.iloc[index]['x'] - head * head_interval + stopper_pos[0],
                                       pcb_data.iloc[index]['y'] + stopper_pos[1]])
-                    mount_angle.append(pcb_data.iloc[index]['r'])
+                    head_angle[head] = pcb_data.iloc[index]['r']
 
                 # 单独计算贴装路径
                 for cntPoints in range(len(mount_pos) - 1):
                     info.place_distance += max(abs(mount_pos[cntPoints][0] - mount_pos[cntPoints + 1][0]),
                                                abs(mount_pos[cntPoints][1] - mount_pos[cntPoints + 1][1]))
 
+                if mount_pos[0][0] < mount_pos[-1][0]:
+                    mount_pos = reversed(mount_pos)
+
                 # 考虑R轴预旋转，补偿同轴角度转动带来的额外贴装用时
-                info.operation_time += head_rotary_time(mount_angle[0])  # 补偿角度转动带来的额外贴装用时
                 info.operation_time += t_nozzle_put * nozzle_put_counter + t_nozzle_pick * nozzle_pick_counter
                 for idx, pos in enumerate(mount_pos):
                     info.operation_time += t_place
-                    move_time = max(axis_moving_time(cur_pos[0] - pos[0], 0), axis_moving_time(cur_pos[1] - pos[1], 1))
+
                     if idx == 0:
+                        move_time = max(axis_moving_time(cur_pos[0] - pos[0], 0),
+                                        axis_moving_time(cur_pos[1] - pos[1], 1))
                         info.round_time += move_time
                     else:
+
+                        cur_head = optimizer_result.head_sequence[cycle][idx]
+                        side_head = cur_head - 1 if cur_head % 2 else cur_head + 1
+                        if optimizer_result.head_sequence[cycle][idx - 1] != side_head:
+                            move_time = max(axis_moving_time(cur_pos[0] - pos[0], 0),
+                                            axis_moving_time(cur_pos[1] - pos[1], 1))
+                        else:
+                            move_time = max(axis_moving_time(cur_pos[0] - pos[0], 0),
+                                            axis_moving_time(cur_pos[1] - pos[1], 1),
+                                            head_rotary_time(head_angle[cur_head] - head_angle[side_head]))
                         info.place_time += move_time
 
                     info.total_distance += max(abs(cur_pos[0] - pos[0]), abs(cur_pos[1] - pos[1]))
@@ -594,5 +629,3 @@ def placement_info_evaluation(component_data, pcb_data, optimizer_result, hinter
     return info
 
 
-
-

base_optimizer/smopt_aggregation.py

@@ -1,12 +1,12 @@
 from base_optimizer.optimizer_common import *
-
+from base_optimizer.smtopt_route import *
 
 def list_range(start, end=None):
     return list(range(start)) if end is None else list(range(start, end))
 
 
 @timer_wrapper
-def optimizer_aggregation(component_data, pcb_data):
+def optimizer_aggregation(component_data, pcb_data, hinter=True):
     # === phase 0: data preparation ===
     M = 1000                                # a sufficient large number
     a, b = 1, 6                             # coefficient
@@ -43,7 +43,7 @@ def optimizer_aggregation(component_data, pcb_data):
 
     # === phase 1: mathematical model solver ===
     mdl = Model('SMT')
-    mdl.setParam('OutputFlag', 0)
+    mdl.setParam('OutputFlag', hinter)
 
     # === Decision Variables ===
     # the number of components of type i that are placed by nozzle type j on placement head k
@@ -104,8 +104,7 @@ def optimizer_aggregation(component_data, pcb_data):
     mdl.setParam("TimeLimit", 100)
 
     mdl.optimize()
-
-    if mdl.Status == GRB.OPTIMAL:
+    if mdl.Status == GRB.OPTIMAL or mdl.Status == GRB.TIME_LIMIT:
         print('total cost = {}'.format(mdl.objval))
 
         # convert cp model solution to standard output
@@ -160,25 +159,9 @@ def optimizer_aggregation(component_data, pcb_data):
         feeder_slot_result = feeder_assignment(component_data, pcb_data, component_result, cycle_result)
 
         # === phase 2: heuristic method ===
-        mount_point_pos = defaultdict(list)
-        for pcb_idx, data in pcb_data.iterrows():
-            part = data['part']
-            part_index = component_data[component_data['part'] == part].index.tolist()[0]
-            mount_point_pos[part_index].append([data['x'], data['y'], pcb_idx])
-
-        for index_ in mount_point_pos.keys():
-            mount_point_pos[index_].sort(key=lambda x: (x[1], x[0]))
-
-        for cycle_idx, _ in enumerate(cycle_result):
-            for _ in range(cycle_result[cycle_idx]):
-                placement_result.append([-1 for _ in range(max_head_index)])
-                for head in range(max_head_index):
-                    if component_result[cycle_idx][head] == -1:
-                        continue
-                    index_ = component_result[cycle_idx][head]
-                    placement_result[-1][head] = mount_point_pos[index_][-1][2]
-                    mount_point_pos[index_].pop()
-                head_sequence.append(dynamic_programming_cycle_path(pcb_data, placement_result[-1], feeder_slot_result[cycle_idx]))
+        placement_result, head_sequence = greedy_level_placing_route_generation(component_data, pcb_data,
+                                                                                component_result, cycle_result,
+                                                                                feeder_slot_result)
 
     else:
         warnings.warn('No solution found!', UserWarning)

base_optimizer/smopt_celldivision.py

@@ -33,7 +33,7 @@ def convert_cell_2_result(pcb_data, component_data, component_cell, population):
 
     wl = [0 for _ in range(max_head_index)]     # workload
 
-    e1, e2, e3 = 1, 0.5, 1. / 6
+    e1, e2, e3 = 1, 2, 1. / 6
 
     component_result, cycle_result, feeder_slot_result = [], [], []
     for index in population:
@@ -101,12 +101,16 @@ def optimizer_celldivision(pcb_data, component_data, hinter=True):
     golden_section = 0.618
 
     # 获取元件元胞
-    point_num = len(pcb_data)
-    component_cell = pd.DataFrame({'index': np.arange(len(component_data)), 'points': np.zeros(len(component_data), dtype=int)})
-    for point_cnt in range(point_num):
-        part = pcb_data.loc[point_cnt, 'part']
-        index = np.where(component_data['part'].values == part)
-        component_cell.loc[index[0], 'points'] += 1
+    feeder_num = sum(component_data['fdn'])
+    component_cell = pd.DataFrame({'index': np.arange(feeder_num), 'points': np.zeros(feeder_num, dtype=int)})
+    cell_index = 0
+    for part_index, data in component_data.iterrows():
+        total_points, division_points = data.points, math.ceil(data.points / data.fdn)
+        for _ in range(data.fdn):
+            component_cell.loc[cell_index, 'index'] = part_index
+            component_cell.loc[cell_index, 'points'] = min(division_points, total_points)
+            total_points -= division_points
+            cell_index += 1
     component_cell = component_cell[~component_cell['points'].isin([0])]
 
     # component_cell.sort_values(by = "points" , inplace = True, ascending = False)

base_optimizer/smopt_hybridgenetic.py

@@ -1,5 +1,5 @@
 from base_optimizer.optimizer_common import *
-
+from base_optimizer.smtopt_route import *
 
 def dynamic_programming_cycle_path(cycle_placement, cycle_points):
     head_sequence = []
@@ -231,8 +231,8 @@ def cal_individual_val(component_nozzle, component_point_pos, designated_nozzle,
 
 def convert_individual_2_result(component_data, component_point_pos, designated_nozzle, pickup_group,
                                 pickup_group_cycle, pair_group, feeder_lane, individual):
+
     component_result, cycle_result, feeder_slot_result = [], [], []
-    placement_result, head_sequence_result = [], []
 
     # === 记录不同元件对应的槽位 ===
     feeder_part_arrange = defaultdict(list)
@@ -273,26 +273,7 @@ def convert_individual_2_result(component_data, component_point_pos, designated_
 
                 pickup_cycle_result[idx][head] -= cycle
 
-    component_point_index = defaultdict(int)
-    for cycle_set in range(len(cycle_result)):
-        for cycle in range(cycle_result[cycle_set]):
-            placement_result.append([-1 for _ in range(max_head_index)])
-            mount_point = [[0, 0] for _ in range(max_head_index)]
-            for head in range(max_head_index):
-                part_index = component_result[cycle_set][head]
-                if part_index == -1:
-                    continue
-
-                part = component_data.iloc[part_index]['part']
-                point_info = component_point_pos[part][component_point_index[part]]
-                
-                placement_result[-1][head] = point_info[2]
-                mount_point[head] = point_info[0:2]
-
-                component_point_index[part] += 1
-            head_sequence_result.append(dynamic_programming_cycle_path(placement_result[-1], mount_point))
-
-    return component_result, cycle_result, feeder_slot_result, placement_result, head_sequence_result
+    return component_result, cycle_result, feeder_slot_result
 
 
 @timer_wrapper
@@ -510,11 +491,6 @@ def optimizer_hybrid_genetic(pcb_data, component_data, hinter=True):
             pop_val.append(val)         # val is related to assembly time
 
         for _ in range(n_generations):
-            # idx = np.argmin(pop_val)
-            # if len(best_pop_val) == 0 or pop_val[idx] < best_pop_val[-1]:
-            #     best_individual = copy.deepcopy(population[idx])
-            # best_pop_val.append(pop_val[idx])
-
             # min-max convert
             max_val = 1.5 * max(pop_val)
             convert_pop_val = list(map(lambda v: max_val - v, pop_val))
@@ -565,6 +541,15 @@ def optimizer_hybrid_genetic(pcb_data, component_data, hinter=True):
             pbar.update(1)
 
     best_individual = population[np.argmin(pop_val)]
+    component_result, cycle_result, feeder_slot_result = convert_individual_2_result(component_data,
+                                                                                     component_point_pos,
+                                                                                     designated_nozzle, pickup_group,
+                                                                                     pickup_group_cycle, pair_group,
+                                                                                     feeder_lane, best_individual)
+    placement_result, head_sequence_result = place_cluster_greedy_route_generation(component_data, pcb_data,
+                                                                                   component_result, cycle_result,
+                                                                                   feeder_slot_result)
+
+
+    return component_result, cycle_result, feeder_slot_result, placement_result, head_sequence_result
 
-    return convert_individual_2_result(component_data, component_point_pos, designated_nozzle, pickup_group,
-                                       pickup_group_cycle, pair_group, feeder_lane, best_individual)

base_optimizer/smopt_mathmodel.py

@@ -1,4 +1,5 @@
 from base_optimizer.optimizer_common import *
+from base_optimizer.smtopt_route import *
 
 
 def head_task_model(component_data, pcb_data, hinter=True):
@@ -6,11 +7,10 @@ def head_task_model(component_data, pcb_data, hinter=True):
     mdl = Model('pick_route')
     mdl.setParam('Seed', 0)
     mdl.setParam('OutputFlag', hinter)          # set whether output the debug information
-    mdl.setParam('TimeLimit', 600)
+    mdl.setParam('TimeLimit', 1000)
 
     H = max_head_index
     I = len(component_data)
-    S = len(component_data)
     K = len(pcb_data)
 
     nozzle_type, component_type = [], []
@@ -29,12 +29,16 @@ def head_task_model(component_data, pcb_data, hinter=True):
     M = 10000
     CompOfNozzle = [[0 for _ in range(J)] for _ in range(I)]    # Compatibility
 
-    component_point = [0 for _ in range(I)]
+    component_point, component_fdn = [0 for _ in range(I)], [0 for _ in range(I)]
+
     for _, data in pcb_data.iterrows():
         idx = component_data[component_data.part == data.part].index.tolist()[0]
         nozzle = component_data.iloc[idx].nz
         CompOfNozzle[idx][nozzle_type.index(nozzle)] = 1
         component_point[idx] += 1
+        component_fdn[idx] = component_data.iloc[idx].fdn
+
+    S = sum(component_fdn)
 
     # objective related
     g = mdl.addVars(list_range(K), vtype=GRB.BINARY)
@@ -94,7 +98,7 @@ def head_task_model(component_data, pcb_data, hinter=True):
         range(-(H - 1) * r, S) for k in range(K))
 
     # feeder related
-    mdl.addConstrs(quicksum(f[s, i] for s in range(S)) <= 1 for i in range(I))
+    mdl.addConstrs(quicksum(f[s, i] for s in range(S)) <= component_fdn[i] for i in range(I))
     mdl.addConstrs(quicksum(f[s, i] for i in range(I)) <= 1 for s in range(S))
     mdl.addConstrs(
         quicksum(x[i, k, h] * y[s, k, h] for h in range(H) for k in range(K)) >= f[s, i] for i in range(I) for s in range(S))
@@ -110,7 +114,7 @@ def head_task_model(component_data, pcb_data, hinter=True):
     # objective
     mdl.setObjective(Fit_cy * quicksum(g[k] for k in range(K)) + Fit_nz * quicksum(
         d[k, h] for h in range(H) for k in range(K)) + Fit_pu * quicksum(
-        e[s, k] for s in range(-(H - 1) * r, S) for k in range(K)) + Fit_mv * head_interval * quicksum(u[k] for k in range(K)),
+        e[s, k] for s in range(-(H - 1) * r, S) for k in range(K)) + Fit_mv * quicksum(u[k] for k in range(K)),
                      GRB.MINIMIZE)
 
     mdl.optimize()
@@ -119,19 +123,21 @@ def head_task_model(component_data, pcb_data, hinter=True):
     for k in range(K):
         if abs(g[k].x) < 1e-6:
             continue
+        component_assign, feeder_slot_assign = [-1 for _ in range(H)], [-1 for _ in range(H)]
 
-        component_result.append([-1 for _ in range(H)])
-        feeder_slot_result.append([-1 for _ in range(H)])
-        cycle_result.append(1)
         for h in range(H):
             for i in range(I):
                 if abs(x[i, k, h].x) > 1e-6:
-                    component_result[-1][h] = i
+                    component_assign[h] = i
 
             for s in range(S):
                 if abs(y[s, k, h].x) > 1e-6:
-                    feeder_slot_result[-1][h] = slot_start + s * interval_ratio - 1
+                    feeder_slot_assign[h] = slot_start + s * interval_ratio - 1
 
+        if sum(component_assign) != -H:
+            component_result.append(component_assign)
+            feeder_slot_result.append(feeder_slot_assign)
+            cycle_result.append(1)
 
     if hinter:
         print(component_result)
@@ -143,7 +149,7 @@ def place_route_model(component_data, pcb_data, component_result, feeder_slot_re
     mdl = Model('place_route')
     mdl.setParam('Seed', 0)
     mdl.setParam('OutputFlag', hinter)  # set whether output the debug information
-    mdl.setParam('TimeLimit', 10)
+    mdl.setParam('TimeLimit', 1000)
 
     component_type = []
     for _, data in component_data.iterrows():
@@ -227,7 +233,7 @@ def place_route_model(component_data, pcb_data, component_result, feeder_slot_re
             else:
                 # there are components on the head
                 mdl.addConstrs(quicksum(w[p, q, k, a] for a in A_from(h) for q in range(P)) + quicksum(
-                    w[q, p, k, a] for a in A_to(h) for q in range(P)) + y[p, k, h] + z[p, k, h] <= 4 *
+                    w[q, p, k, a] for a in A_to(h) for q in range(P)) + y[p, k, h] + z[p, k, h] <= 2 *
                                CompOfPoint[component_result[k][h]][p] for p in range(P))
 
     # each head corresponds to a maximum of one point in each cycle
@@ -362,7 +368,9 @@ def place_route_model(component_data, pcb_data, component_result, feeder_slot_re
 def optimizer_mathmodel(component_data, pcb_data, hinter=True):
 
     component_result, cycle_result, feeder_slot_result = head_task_model(component_data, pcb_data, hinter)
-    placement_result, head_sequence = place_route_model(component_data, pcb_data, component_result, feeder_slot_result)
-    # placement_result, head_sequence = greedy_placement_route_generation(component_data, pcb_data, component_result,
-    #                                                                     cycle_result)
+    # placement_result, head_sequence = place_route_model(component_data, pcb_data, component_result, feeder_slot_result)
+    placement_result, head_sequence = place_allocate_sequence_route_generation(component_data, pcb_data,
+                                                                               component_result, cycle_result,
+                                                                               feeder_slot_result)
+
     return component_result, cycle_result, feeder_slot_result, placement_result, head_sequence

base_optimizer/smopt_twophase.py

@@ -1,4 +1,6 @@
 from base_optimizer.optimizer_common import *
+from base_optimizer.smtopt_route import *
+from base_optimizer.result_analysis import *
 
 
 def list_range(start, end=None):
@@ -81,8 +83,9 @@ def gurobi_optimizer(pcb_data, component_data, feeder_data, reduction=True, part
             changetime = mdl.cbGet(GRB.Callback.RUNTIME)
             nonlocal pre_objbst, pre_changetime
 #            condition: value change
+            if abs(objbst - 1e+100) > 1:    # 避免未找到可行解提前退出
                 if pre_objbst and abs(pre_objbst - objbst) < 1e-3:
-                if pre_changetime and changetime - pre_changetime > 100 * (1 - objbnd / objbst):
+                    if pre_changetime and changetime - pre_changetime > 90 * (1 - objbnd / objbst):
                         mdl.terminate()
                 else:
                     pre_changetime = changetime
@@ -147,7 +150,7 @@ def gurobi_optimizer(pcb_data, component_data, feeder_data, reduction=True, part
         level += 1
 
     L = len(cycle_assignment) if partition else len(pcb_data)
-    S = ratio * sum(component_feeder.values()) * 2 if len(feeder_data) == 0 else arg_slot_rng[-1] - arg_slot_rng[0] + 1 # the available feeder num
+    S = ratio * sum(component_feeder.values()) if len(feeder_data) == 0 else arg_slot_rng[-1] - arg_slot_rng[0] + 1 # the available feeder num
     M = len(pcb_data)  # a sufficiently large number (number of placement points)
     HC = [[0 for _ in range(J)] for _ in range(I)]
     for i in range(I):
@@ -157,12 +160,12 @@ def gurobi_optimizer(pcb_data, component_data, feeder_data, reduction=True, part
     mdl = Model('SMT')
     mdl.setParam('Seed', 0)
     mdl.setParam('OutputFlag', hinter)  # set whether output the debug information
-    mdl.setParam('TimeLimit', 3600 * 3)
+    mdl.setParam('TimeLimit', 3600)
     mdl.setParam('PoolSearchMode', 2)
-    mdl.setParam('PoolSolutions', 3e2)
+    mdl.setParam('PoolSolutions', 100)
     mdl.setParam('PoolGap', 1e-4)
     # mdl.setParam('MIPFocus', 2)
-    # mdl.setParam("Heuristics", 0.5)
+    mdl.setParam("Heuristics", 0.5)
 
     # Use only if other methods, including exploring the tree with the default settings, do not yield a viable solution
     # mdl.setParam("ZeroObjNodes", 100)
@@ -178,7 +181,6 @@ def gurobi_optimizer(pcb_data, component_data, feeder_data, reduction=True, part
         c[i, h, l] <= component_list[cpidx_2_part[i]] for i in range(I) for h in range(max_head_index) for l in
         range(L))
 
-    # todo: the condition for upper limits of feeders exceed 1
     f = {}
     for i in range(I):
         if i not in part_feederbase.keys():
@@ -227,7 +229,7 @@ def gurobi_optimizer(pcb_data, component_data, feeder_data, reduction=True, part
                 continue
 
             part_feederbase[part_2_cpidx[part]] = slot
-            # f[slot, part_2_cpidx[part]].Start = 1
+            f[slot, part_2_cpidx[part]].Start = 1
             slot += 1
 
         for idx, cycle in enumerate(cycle_index):
@@ -334,13 +336,12 @@ def gurobi_optimizer(pcb_data, component_data, feeder_data, reduction=True, part
         in range(L))
 
     if reduction:
-        # mdl.addConstrs(WL[l] >= WL[l + 1] for l in range(L - 1))
+        mdl.addConstrs(WL[l] >= WL[l + 1] for l in range(L - 1))
         mdl.addConstr(quicksum(WL[l] for l in range(L)) <= sum(cycle_assignment))
         mdl.addConstr(quicksum(WL[l] for l in range(L)) >= math.ceil(len(pcb_data) / max_head_index))
         mdl.addConstrs(quicksum(z[j, h, l] for j in range(J) for h in range(max_head_index)) >= quicksum(
             z[j, h, l + 1] for j in range(J) for h in range(max_head_index)) for l in range(L - 1))
 
-
     mdl.addConstrs(y[i, h, l] <= WL[l] for i in range(I) for h in range(max_head_index) for l in range(L))
     mdl.addConstrs(v[s, h, l] <= WL[l] for s in range(S) for h in range(max_head_index) for l in range(L))
 
@@ -365,125 +366,126 @@ def gurobi_optimizer(pcb_data, component_data, feeder_data, reduction=True, part
     # mdl.optimize()
 
     # === result generation ===
-    nozzle_assign, component_assign = [], []
-    feeder_assign, cycle_assign = [], []
+    opt_res_list = defaultdict(OptResult)
     if mdl.Status == GRB.OPTIMAL or mdl.Status == GRB.INTERRUPTED or mdl.Status == GRB.TIME_LIMIT:
         # === selection from solution pool ===
-        component_pos, component_avg_pos = defaultdict(list), defaultdict(list)
+        component_pos = defaultdict(list[Point])
         for _, data in pcb_data.iterrows():
             component_index = component_data[component_data.part == data.part].index.tolist()[0]
-            component_pos[component_index].append([data.x, data.y])
+            component_pos[component_index].append(Point(data.x, data.y))
 
-        for i in component_pos.keys():
-            component_pos[i] = sorted(component_pos[i], key=lambda pos: (pos[0], pos[1]))
-            component_avg_pos[i] = [sum(map(lambda pos: pos[0], component_pos[i])) / len(component_pos[i]),
-                                    sum(map(lambda pos: pos[1], component_pos[i])) / len(component_pos[i])]
+        for part in component_pos.keys():
+            component_pos[part] = sorted(component_pos[part], key=lambda pos: (pos.x, pos.y))
 
-        min_dist, solution_number = None, -1
+        min_dist, solution_number = None, 0
         for sol_counter in range(mdl.SolCount):
-            nozzle_assign, component_assign = [], []
-            feeder_assign, cycle_assign = [], []
-
             mdl.Params.SolutionNumber = sol_counter
-            pos_counter = defaultdict(int)
-
-            dist = 0
-            cycle_placement, cycle_points = defaultdict(list), defaultdict(list)
+            opt_res = OptResult()
+            # == 转换标准的贴装头分配的解 ===
             for l in range(L):
                 if abs(WL[l].Xn) <= 1e-4:
                     continue
-                cycle_placement[l], cycle_points[l] = [-1] * max_head_index, [None] * max_head_index
+                opt_res.cycle_assign.append(round(WL[l].Xn))
+                opt_res.component_assign.append([-1] * max_head_index)
+                opt_res.feeder_slot_assign.append([-1] * max_head_index)
 
                 for h in range(max_head_index):
-                for l in range(L):
-                    if abs(WL[l].Xn) <= 1e-4:
-                        continue
-
-                    pos_list = []
                     for i in range(I):
                         if abs(y[i, h, l].Xn) <= 1e-4:
                             continue
+                        opt_res.component_assign[-1][h] = i
 
-                        for _ in range(round(WL[l].Xn)):
-                            pos_list.append(component_pos[i][pos_counter[i]])
-                            pos_counter[i] += 1
-
-                        cycle_placement[l][h] = i
-                        cycle_points[l][h] = [sum(map(lambda pos: pos[0], pos_list)) / len(pos_list),
-                                              sum(map(lambda pos: pos[1], pos_list)) / len(pos_list)]
-            for l in range(L):
-                if abs(WL[l].Xn) <= 1e-4:
-                    continue
-
-            if min_dist is None or dist < min_dist:
-                min_dist = dist
-                solution_number = sol_counter
-
-        mdl.Params.SolutionNumber = solution_number
-        # === 更新吸嘴、元件、周期数优化结果 ===
-        for l in range(L):
-            nozzle_assign.append([-1 for _ in range(max_head_index)])
-            component_assign.append([-1 for _ in range(max_head_index)])
-            feeder_assign.append([-1 for _ in range(max_head_index)])
-
-            cycle_assign.append(round(WL[l].Xn))
-            if abs(WL[l].Xn) <= 1e-4:
-                continue
-
-            for h in range(max_head_index):
-                for i in range(I):
-                    if abs(y[i, h, l].Xn - 1) < 1e-4:
-                        component_assign[-1][h] = i
-
-                        for j in range(J):
-                            if HC[i][j]:
-                                nozzle_assign[-1][h] = j
                     for s in range(S):
-                    if abs(v[s, h, l].Xn - 1) < 1e-4 and component_assign[l][h] != -1:
-                        feeder_assign[l][h] = s
+                        if abs(v[s, h, l].Xn - 1) < 1e-4 and opt_res.component_assign[-1][h] != -1:
+                            opt_res.feeder_slot_assign[-1][h] = s
 
-        # === 更新供料器分配结果 ==
-        component_head = defaultdict(int)
-        for i in range(I):
-            cycle_num = 0
-            for l, component_cycle in enumerate(component_assign):
-                for head, component in enumerate(component_cycle):
-                    if component == i:
-                        component_head[i] += cycle_assign[l] * head
-                        cycle_num += cycle_assign[l]
-            component_head[i] /= cycle_num      # 不同元件的加权拾取贴装头
-
-        average_pos = 0
-        for _, data in pcb_data.iterrows():
-            average_pos += (data.x - component_head[part_2_cpidx[data.part]] * head_interval)
-
-        average_pos /= len(pcb_data)    # 实际贴装位置的加权平均
-        average_slot = 0
-        for l in range(L):
-            if abs(WL[l].Xn) <= 1e-4:
+            # 根据贴装头位置，转换供料器槽位
+            cp_avg_head, cp_sum_cycle = defaultdict(float), defaultdict(int)
+            for cycle, component_assign in enumerate(opt_res.component_assign):
+                for head, part in enumerate(component_assign):
+                    if part == -1:
                         continue
-            min_slot, max_slot = None, None
+                    cp_avg_head[part] += opt_res.cycle_assign[cycle] * head
+                    cp_sum_cycle[part] += opt_res.cycle_assign[cycle]
+
+            for part, head in cp_avg_head.items():
+                cp_avg_head[part] = head / cp_sum_cycle[part]
+
+            avg_position = sum([data.x - cp_avg_head[part_2_cpidx[data.part]] * head_interval for _, data in
+                                pcb_data.iterrows()]) / len(pcb_data)
+            avg_slot = 0
+            D_PU, D_PL, D_BW, D_FW = 0, 0, 0, 0
+            for cycle, slots in enumerate(opt_res.feeder_slot_assign):
+                min_slot, max_slot = max_slot_index, 0
+                for head, slot in enumerate(slots):
+                    if slot == -1:
+                        continue
+                    min_slot = min(min_slot, slot - head * ratio)
+                    max_slot = max(max_slot, slot - head * ratio)
+                avg_slot += (max_slot - min_slot) * opt_res.cycle_assign[cycle]
+                D_PU += (max_slot - min_slot) * slot_interval * opt_res.cycle_assign[cycle]     # 拾取路径
+
+            avg_slot /= sum(opt_res.cycle_assign)
+            start_slot = round((avg_position + stopper_pos[0] - slotf1_pos[0]) / slot_interval + avg_slot / 2) + 1
+
+            for cycle in range(len(opt_res.feeder_slot_assign)):
                 for head in range(max_head_index):
-                if abs(WL[l].Xn) <= 1e-4 or feeder_assign[l][head] == -1:
+                    if (slot := opt_res.feeder_slot_assign[cycle][head]) == -1:
                         continue
-                slot = feeder_assign[l][head] - head * ratio
-                if min_slot is None or slot < min_slot:
-                    min_slot = slot
-                if max_slot is None or slot > max_slot:
-                    max_slot = slot
-            average_slot += (max_slot - min_slot) * cycle_assign[l]
-        average_slot /= sum(cycle_assign)
-        start_slot = round((average_pos + stopper_pos[0] - slotf1_pos[0]) / slot_interval + average_slot / 2) + 1
+                    opt_res.feeder_slot_assign[cycle][head] = start_slot + slot * (2 if ratio == 1 else 1)
 
-        for l in range(L):
-            if abs(WL[l].Xn) <= 1e-4:
+            component_pos_counter = defaultdict(int)
+            cycle_place_pos = defaultdict(list[Point])
+            for head in range(max_head_index):
+                for cycle in range(len(opt_res.cycle_assign)):
+                    if (part := opt_res.component_assign[cycle][head]) == -1:
                         continue
 
-            for h in range(max_head_index):
-                for s in range(S):
-                    if abs(v[s, h, l].Xn - 1) < 1e-4 and component_assign[l][h] != -1:
-                        feeder_assign[l][h] = start_slot + s * (2 if ratio == 1 else 1)
+                    avg_place_pos = Point(0, 0, _h=head)
+                    for counter in range(round(opt_res.cycle_assign[cycle])):
+                        avg_place_pos.x = (1 - 1.0 / (counter + 1)) * avg_place_pos.x + (
+                                component_pos[part][component_pos_counter[part]].x - head * head_interval) / (
+                                                  counter + 1)
+                        avg_place_pos.y = (1 - 1.0 / (counter + 1)) * avg_place_pos.y + component_pos[part][
+                            component_pos_counter[part]].y / (counter + 1)
+                        component_pos_counter[part] += 1
+                    avg_place_pos.x += stopper_pos[0]
+                    avg_place_pos.y += stopper_pos[1]
+                    cycle_place_pos[cycle].append(avg_place_pos)
 
+            for cycle in range(len(opt_res.cycle_assign)):
+                min_slot, max_slot = max_slot_index, 0
+                for head in range(max_head_index):
+                    if (slot := opt_res.feeder_slot_assign[cycle][head]) == -1:
+                        continue
+                    min_slot = min(min_slot, slot - head * interval_ratio)
+                    max_slot = max(max_slot, slot - head * interval_ratio)
+                # cycle_place_pos[cycle] = sorted(cycle_place_pos[cycle], key=lambda pt: pt.x)
+
+                pick_pos = Point(slotf1_pos[0] + (min_slot + max_slot) / 2 * slot_interval, slotf1_pos[1])
+                _, seq = dynamic_programming_cycle_route(cycle_place_pos[cycle], pick_pos)
+                head_position = [Point(0, 0) for _ in range(max_head_index)]
+                for point in cycle_place_pos[cycle]:
+                    head_position[point.h] = point
+
+                for idx in range(len(seq) - 1):
+                    h1, h2 = seq[idx], seq[idx + 1]
+                    D_PL += max(abs(head_position[h1].x - head_position[h2].x),
+                                abs(head_position[h1].y - head_position[h2].y)) * opt_res.cycle_assign[cycle]
+
+
+            # opt_res.placement_assign, opt_res.head_sequence = scan_based_placement_route_generation(component_data,
+            #                                                                                         pcb_data,
+            #                                                                                         opt_res.component_assign,
+            #                                                                                         opt_res.cycle_assign,
+            #                                                                                         opt_res.feeder_slot_assign,
+            #                                                                                         hinter=False)
+            # info = placement_info_evaluation(component_data, pcb_data, opt_res)
+            # print(f'{info.place_distance + info.pickup_distance: .3f}\t{D_PL + D_PU: .3f}')
+            opt_res_list[sol_counter] = opt_res
+
+    solution_number = 0
+    mdl.Params.SolutionNumber = 0
     if hinter:
         print('total cost = {}'.format(mdl.objval))
         print('cycle = {}, nozzle change = {}, pick up = {}'.format(quicksum(WL[l].Xn for l in range(L)), quicksum(
@@ -496,9 +498,10 @@ def gurobi_optimizer(pcb_data, component_data, feeder_data, reduction=True, part
 
         print('')
         print('result')
-            print('nozzle assignment: ', nozzle_assign)
-            print('component assignment: ', component_assign)
-            print('feeder assignment: ', feeder_assign)
-            print('cycle assignment: ', cycle_assign)
-    return component_assign, feeder_assign, cycle_assign
+        print('component assignment: ', opt_res_list[solution_number].component_assign)
+        print('feeder assignment: ', opt_res_list[solution_number].feeder_slot_assign)
+        print('cycle assignment: ', opt_res_list[solution_number].cycle_assign)
+
+    return opt_res_list[solution_number].component_assign, opt_res_list[solution_number].feeder_slot_assign, \
+           opt_res_list[solution_number].cycle_assign
 

dataloader.py

@@ -161,8 +161,27 @@ def load_data(filename: str, load_feeder=False, auto_register=True):
             feeder_data = pd.DataFrame(columns=feeder_columns)
     else:
         feeder_data = pd.DataFrame(columns=feeder_columns)
+
+    for idx, data in feeder_data.iterrows():
+        feeder_data.at[idx, 'slot'] = int(data['slot'][1:])
+
     feeder_data.sort_values(by='slot', ascending=True, inplace=True, ignore_index=True)
 
+
+    feeder_data_check = defaultdict(str)
+    for _, data in feeder_data.iterrows():
+        feeder_data_check[data.slot] = data.part
+
+    for machine_index in range(machine_num):
+        for idx, data in pcb_data[machine_index].iterrows():
+            slot, part = data.fdr.split(' ')
+            slot = int(slot[1:])
+            if feeder_data_check[slot] == '':
+                feeder_data_check[slot] = part
+            if feeder_data_check[slot] != part:
+                warning_info = f'conflict feeder registration PCB: {data.fdr}, BASE: F{slot} {feeder_data_check[slot]}'
+                # warnings.warn(warning_info, UserWarning)
+
     return pcb_data, component_data, feeder_data
 
 

estimator.py

@@ -6,7 +6,7 @@ def exact_assembly_time(pcb_data, component_data):
     feeder_data = pd.DataFrame(columns=['slot', 'part'])
     component_result, cycle_result, feeder_slot_result = feeder_priority_assignment(component_data, pcb_data,
                                                                                     feeder_data, hinter=False)
-    placement_result, head_sequence_result = greedy_placement_route_generation(component_data, pcb_data,
+    placement_result, head_sequence_result = place_allocate_sequence_route_generation(component_data, pcb_data,
                                                                                       component_result, cycle_result,
                                                                                       feeder_slot_result, hinter=False)
     opt_res = OptResult(component_result, cycle_result, feeder_slot_result, placement_result, head_sequence_result)
@@ -19,7 +19,7 @@ def error_info(pred_val, real_val, type='train'):
     absolute_error = np.array([])
     for idx, (t1, t2) in enumerate(np.nditer([pred_val, real_val])):
         absolute_error = np.append(absolute_error, abs(t1 - t2) / (t2 + 1e-10) * 100)
-        if absolute_error[-1] > 15:
+        if absolute_error[-1] > 10:
             print(f'\033[0;31;31midx: {idx + 1: d}, net: {t1: .3f},  real: {t2: .3f}, '
                   f'gap: {absolute_error[-1]: .3f}\033[0m')
 
@@ -561,6 +561,8 @@ class MetricEstimator(Estimator):
     def training(self, params):
         x_fit, y_fit = data_mgr.metric('opt/' + params.train_file)
         self.lr.fit(x_fit, y_fit)
+        y_predict = self.lr.predict(x_fit)
+        error_info(y_fit, y_predict, 'train')
         print(self.lr.coef_)
 
     def testing(self, params):
@@ -578,7 +580,7 @@ if __name__ == '__main__':
 
     parser = argparse.ArgumentParser(description='network training implementation')
     # parser.add_argument('--train', default=True, type=bool, help='determine whether training the network')
-    parser.add_argument('--save', default=True, type=bool,
+    parser.add_argument('--save', default=False, type=bool,
                         help='determine whether saving the parameters of network, linear regression model, etc.')
     parser.add_argument('--overwrite', default=False, type=bool,
                         help='determine whether overwriting the training and testing data')
@@ -588,7 +590,7 @@ if __name__ == '__main__':
     parser.add_argument('--batch_size', default=2000, type=int, help='size of training batch')
     parser.add_argument('--lr', default=1e-5, type=float, help='learning rate for the network')
     parser.add_argument('--model', default='neural-network', help='method for assembly time estimation')
-    parser.add_argument('--machine_optimizer', default='feeder-scan', type=str, help='optimizer for single machine')
+    parser.add_argument('--machine_optimizer', default='feeder-priority', type=str, help='optimizer for single machine')
     params = parser.parse_args()
 
     data_mgr = DataMgr()
@@ -606,13 +608,13 @@ if __name__ == '__main__':
                     # data_mgr.remover()                                # remove the last saved data
                     # data_mgr.saver('data/' + file_name, pcb_data)     # save new data
 
-                    info = base_optimizer(1, pcb_data, component_data, pd.DataFrame(columns=['slot', 'part', 'arg']),
-                                          params, hinter=True)
+                    info = base_optimizer(1, pcb_data, component_data, pd.DataFrame(columns=['slot', 'part']), params,
+                                          hinter=True)
 
                     data_mgr.recorder(f, info, pcb_data, component_data)
             f.close()
 
-    estimator = MetricEstimator()
+    estimator = NeuralEstimator()
 
     estimator.training(params)
     estimator.testing(params)

generator.py

@@ -73,9 +73,9 @@ class DataMgr:
         lineinfo += '\t' + '{:.3f}'.format(pcb_data['x'].max() - pcb_data['x'].min()) + '\t' + '{:.3f}'.format(
             pcb_data['y'].max() - pcb_data['y'].min())
 
-        # part_position = defaultdict(list)
-        # for _, data in pcb_data.iterrows():
-        #     part_position[data['part']].append([data['x'], data['y']])
+        part_position = defaultdict(list)
+        for _, data in pcb_data.iterrows():
+            part_position[data['part']].append([data['x'], data['y']])
 
         point_counter, component_counter = 0, 0
         nozzle_type = set()
@@ -94,8 +94,8 @@ class DataMgr:
             if data.points == 0:
                 continue
             lineinfo += '\t' + data.part + '\t' + data.nz + '\t' + str(data.points)
-            # lineinfo += '\t' + '{:.3f}'.format(np.ptp([pos[0] for pos in part_position[data.part]]))
-            # lineinfo += '\t' + '{:.3f}'.format(np.ptp([pos[1] for pos in part_position[data.part]]))
+            lineinfo += '\t' + '{:.3f}'.format(np.ptp([pos[0] for pos in part_position[data.part]]))
+            lineinfo += '\t' + '{:.3f}'.format(np.ptp([pos[1] for pos in part_position[data.part]]))
 
         lineinfo += '\n'
         file_handle.write(lineinfo)
@@ -173,7 +173,7 @@ class DataMgr:
         return data
 
     def heuristic_objective(self, cp_points, cp_nozzle):
-        if len(cp_points.keys()) == 0:
+        if len(cp_points.keys()) or sum(cp_points.values()) == 0:
             return 0, 0, 0, 0
         nozzle_heads, nozzle_points = defaultdict(int), defaultdict(int)
         for idx, points in cp_points.items():
@@ -417,18 +417,18 @@ class DataMgr:
                 # assembly time data
                 time_data.append(float(items[0]))
 
-                cp_data.append([cp_points, cp_nozzle, board_width, board_height])
+                cp_data.append([cp_points, cp_nozzle, board_width, board_height, cycle, nozzle_change_data, pickup_data])
 
                 line = file.readline()
 
         if data_filter:
-            cph_data = [point_data[idx] / time_data[idx] * 3600 for idx in range(len(time_data))]
+            # cph_data = [point_data[idx] / time_data[idx] * 3600 for idx in range(len(time_data))]
 
-            w_quart = 0.6
-            Q1, Q3 = np.percentile(np.array(cph_data), 25), np.percentile(np.array(cph_data), 75)
+            w_quart = 0.3
+            Q1, Q3 = np.percentile(np.array(time_data), 25), np.percentile(np.array(time_data), 75)
 
-            indices = [i for i in range(len(cph_data)) if
-                       Q1 - w_quart * (Q3 - Q1) <= cph_data[i] <= Q3 + w_quart * (Q3 - Q1)]
+            indices = [i for i in range(len(time_data)) if
+                       Q1 - w_quart * (Q3 - Q1) <= time_data[i] <= Q3 + w_quart * (Q3 - Q1)]
 
             filter_cp_data, filter_time_data = [], []
             for idx in indices:
@@ -465,8 +465,8 @@ class DataMgr:
 
     def neural_encode(self, input_data):
         train_data = []
-        for cp_points, cp_nozzle, board_width, board_height in input_data:
-            train_data.append(self.encode(cp_points, cp_nozzle, board_width, board_height))
+        for cp_points, cp_nozzle, board_width, board_height, cy, nz, pu in input_data:
+            train_data.append(self.encode(cp_points, cp_nozzle, board_width, board_height, cy, nz, pu))
         return train_data
 
     def get_feature(self):

lineopt_hyperheuristic.py

@@ -278,7 +278,7 @@ def cal_individual_val(heuristic_map, cp_index, cp_points, cp_nozzle, cp_feeders
 def line_optimizer_hyperheuristic(component_data, pcb_data, machine_number):
     heuristic_map = {
         'p': LeastPoints,
-        'n': LeastNzChange,
+        'n': LeastNzTypes,
         'c': LeastCpTypes,
         'r': LeastCpNzRatio,
         'k': LeastCycle,
@@ -411,12 +411,13 @@ def line_optimizer_hyperheuristic(component_data, pcb_data, machine_number):
                     best_component_list = component_list.copy()
 
     machine_cp_points = convert_assignment_result(heuristic_map, cp_index, cp_points, cp_nozzle, cp_feeders,
-                                                  best_component_list, best_heuristic_list, machine_number)
+                                                  best_component_list, best_heuristic_list, machine_number, is_opt=True)
 
     assignment_result = [[0 for _ in range(len(component_data))] for _ in range(machine_number)]
     for machine_idx in range(machine_number):
         for idx in machine_cp_points[machine_idx]:
             assignment_result[machine_idx][cp_index[idx]] += cp_points[idx]
+
     return assignment_result
 
 

lineopt_model.py

@@ -7,7 +7,7 @@ def line_optimizer_model(component_data, pcb_data, machine_num, hinter=True):
     mdl = Model('pcb assembly line optimizer')
     mdl.setParam('Seed', 0)
     mdl.setParam('OutputFlag', hinter)  # set whether output the debug information
-    mdl.setParam('TimeLimit', 600 * 3)
+    mdl.setParam('TimeLimit', 1000)
 
     nozzle_type, component_type = [], []
     for _, data in component_data.iterrows():
@@ -22,9 +22,10 @@ def line_optimizer_model(component_data, pcb_data, machine_num, hinter=True):
 
     H = max_head_index
     I = len(component_data)
-    S = min(len(component_data) * ratio, 60)
-    K = math.ceil(len(pcb_data) * 1.0 / H / M) + 1
-    # K = 3
+    S = sum([data.fdn * ratio for _, data in component_data.iterrows()])
+    K = math.ceil(len(pcb_data) * 1.0 / M) + 1
+
+    # K = len(pcb_data)
     CompOfNozzle = [[0 for _ in range(J)] for _ in range(I)]  # Compatibility
 
     component_point = [0 for _ in range(I)]
@@ -35,15 +36,17 @@ def line_optimizer_model(component_data, pcb_data, machine_num, hinter=True):
 
     # objective related
     g = mdl.addVars(list_range(K), list_range(M), vtype=GRB.BINARY)
-    d = mdl.addVars(list_range(K), list_range(H), list_range(M), lb=0, vtype=GRB.CONTINUOUS)
+    d = mdl.addVars(list_range(K), list_range(H), list_range(M), vtype=GRB.INTEGER)
     u = mdl.addVars(list_range(I), list_range(K), list_range(H), list_range(M), vtype=GRB.BINARY)
     v = mdl.addVars(list_range(S), list_range(K), list_range(H), list_range(M), vtype=GRB.BINARY)
-    d_plus = mdl.addVars(list_range(J), list_range(K), list_range(H), list_range(M), lb=0, vtype=GRB.CONTINUOUS)
-    d_minus = mdl.addVars(list_range(J), list_range(K), list_range(H), list_range(M), lb=0, vtype=GRB.CONTINUOUS)
-    w = mdl.addVars(list_range(K), list_range(M), vtype=GRB.CONTINUOUS)
+    w = mdl.addVars(list_range(J), list_range(K), list_range(H), list_range(M), vtype=GRB.BINARY)
+    d_plus = mdl.addVars(list_range(J), list_range(K), list_range(H), list_range(M), vtype=GRB.CONTINUOUS)
+    d_minus = mdl.addVars(list_range(J), list_range(K), list_range(H), list_range(M), vtype=GRB.CONTINUOUS)
+    z = mdl.addVars(list_range(K), list_range(M), vtype=GRB.INTEGER)
 
     e = mdl.addVars(list_range(-(H - 1) * ratio, S), list_range(K), list_range(M), vtype=GRB.BINARY)
     f = mdl.addVars(list_range(S), list_range(I), list_range(M), vtype=GRB.BINARY, name='')
+    t = mdl.addVars(list_range(M), lb=0, ub=N, vtype=GRB.CONTINUOUS)
     obj = mdl.addVar(lb=0, ub=N, vtype=GRB.CONTINUOUS)
 
     mdl.addConstrs(g[k, m] >= g[k + 1, m] for k in range(K - 1) for m in range(M))
@@ -52,24 +55,26 @@ def line_optimizer_model(component_data, pcb_data, machine_num, hinter=True):
         quicksum(u[i, k, h, m] for i in range(I)) <= g[k, m] for k in range(K) for h in range(H) for m in range(M))
 
     # nozzle no more than 1 for head h and cycle k
-    mdl.addConstrs(quicksum(CompOfNozzle[i][j] * u[i, k, h, m] for i in range(I) for j in range(J)) <= 1 for k
-                   in range(K) for h in range(H) for m in range(M))
+    mdl.addConstrs(quicksum(w[j, k, h, m] for j in range(J)) <= 1 for k in range(K) for h in range(H) for m in range(M))
 
     # work completion
     mdl.addConstrs(
         quicksum(u[i, k, h, m] for k in range(K) for h in range(H) for m in range(M)) == component_point[i] for i in
         range(I))
+
     # nozzle change
-    mdl.addConstrs(quicksum(CompOfNozzle[i][j] * u[i, k, h, m] for i in range(I)) - quicksum(
-        CompOfNozzle[i][j] * u[i, k + 1, h, m] for i in range(I)) == d_plus[j, k, h, m] - d_minus[j, k, h, m] for k in
+    mdl.addConstrs(
+        u[i, k, h, m] <= quicksum(CompOfNozzle[i][j] * w[j, k, h, m] for j in range(J)) for i in range(I) for k in
+        range(K) for h in range(H) for m in range(M))
+
+    mdl.addConstrs(w[j, k, h, m] - w[j, k + 1, h, m] == d_plus[j, k, h, m] - d_minus[j, k, h, m] for k in
                    range(K - 1) for j in range(J) for h in range(H) for m in range(M))
 
-    mdl.addConstrs(quicksum(CompOfNozzle[i][j] * u[i, K - 1, h, m] for i in range(I)) - quicksum(
-        CompOfNozzle[i][j] * u[i, 0, h, m] for i in range(I)) == d_plus[j, K - 1, h, m] - d_minus[j, K - 1, h, m] for j
-                   in range(J) for h in range(H) for m in range(M))
+    mdl.addConstrs(w[j, 0, h, m] - w[j, K - 1, h, m] == d_plus[j, K - 1, h, m] - d_minus[j, K - 1, h, m]
+                   for j in range(J) for h in range(H) for m in range(M))
 
-    mdl.addConstrs(2 * d[k, h, m] == quicksum(d_plus[j, k, h, m] for j in range(J)) + quicksum(
-        d_minus[j, k, h, m] for j in range(J)) - 1 for k in range(K) for h in range(H) for m in range(M))
+    mdl.addConstrs(d[k, h, m] == quicksum(d_plus[j, k, h, m] for j in range(J)) + quicksum(
+        d_minus[j, k, h, m] for j in range(J)) for k in range(K) for h in range(H) for m in range(M))
 
     # simultaneous pick
     for s in range(-(H - 1) * ratio, S):
@@ -99,25 +104,29 @@ def line_optimizer_model(component_data, pcb_data, machine_num, hinter=True):
         for s in range(S) for m in range(M))
 
     # pickup movement
-    mdl.addConstrs(w[k, m] >= s1 * e[s1, k, m] - s2 * e[s2, k, m] + N * (e[s1, k, m] + e[s2, k, m] - 2) for s1 in
+    mdl.addConstrs(z[k, m] >= s1 * e[s1, k, m] - s2 * e[s2, k, m] + N * (e[s1, k, m] + e[s2, k, m] - 2) for s1 in
                    range(-(H - 1) * ratio, S) for s2 in range(-(H - 1) * ratio, S) for k in range(K) for m in range(M))
 
     # objective
-    mdl.addConstrs(obj >= Fit_cy * quicksum(g[k, m] for k in range(K)) + Fit_nz * 2 * quicksum(
+    mdl.addConstrs(t[m] == Fit_cy * quicksum(g[k, m] for k in range(K)) + Fit_nz * quicksum(
                         d[k, h, m] for h in range(H) for k in range(K)) + Fit_pu * quicksum(
                         e[s, k, m] for s in range(-(H - 1) * ratio, S) for k in range(K)) + Fit_pl * quicksum(
-        u[i, k, h, m] for i in range(I) for k in range(K) for h in range(H)) + Fit_mv * head_interval * quicksum(
-        w[k, m] for k in range(K)) for m in range(M))
+                        u[i, k, h, m] for i in range(I) for k in range(K) for h in range(H)) + Fit_mv * quicksum(
+                        z[k, m] for k in range(K)) for m in range(M))
+    for m in range(M - 1):
+        mdl.addConstr(t[m] >= t[m + 1])
+
+    mdl.addConstrs(obj >= t[m] for m in range(M))
 
     mdl.setObjective(obj, GRB.MINIMIZE)
-
     mdl.optimize()
+
     for m in range(M):
         print(f'machine {m} : cycle : {sum(g[k, m].x for k in range(K))}, '
               f'nozzle change : {sum(d[k, h, m].x for h in range(H) for k in range(K))}, '
               f'pick up : {sum(e[s, k, m].x for s in range(-(H - 1) * ratio, S) for k in range(K))}, '
               f'placement : {sum(u[i, k, h, m].x for i in range(I) for k in range(K) for h in range(H))}, '
-              f'pick movement : {sum(w[k, m].x for k in range(K))}')
+              f'pick movement : {sum(z[k, m].x for k in range(K))}')
 
     pcb_part_indices = defaultdict(list)
     for idx, data in pcb_data.iterrows():
@@ -162,19 +171,22 @@ def line_optimizer_model(component_data, pcb_data, machine_num, hinter=True):
 
         average_pos = round(
             (sum(head_place_pos) / len(head_place_pos) + stopper_pos[0] - slotf1_pos[0] + 1) / slot_interval)
-        print(f'average_pos: {average_pos}')
+
         for k in range(len(feeder_slot_result)):
             for h in range(H):
                 if feeder_slot_result[k][h] == -1:
                     continue
                 feeder_slot_result[k][h] = feeder_slot_result[k][h] * 2 + average_pos
 
-        placement_result, head_sequence = greedy_placement_route_generation(partial_component_data, partial_pcb_data,
+        placement_result, head_sequence = place_allocate_sequence_route_generation(partial_component_data,
+                                                                                   partial_pcb_data,
                                                                                    component_result, cycle_result,
                                                                                    feeder_slot_result, hinter=False)
-        print('----- Placement machine ' + str(m + 1) + ' ----- ')
+
         opt_res = OptResult(component_result, cycle_result, feeder_slot_result, placement_result, head_sequence)
-        info = placement_info_evaluation(partial_component_data, partial_pcb_data, opt_res, hinter=False)
+        info = placement_info_evaluation(partial_component_data, partial_pcb_data, opt_res, hinter=hinter)
+        if hinter:
+            print('----- Placement machine ' + str(m + 1) + ' ----- ')
             optimization_assign_result(partial_component_data, partial_pcb_data, opt_res, nozzle_hinter=True,
                                        component_hinter=True, feeder_hinter=True)
             info.print()

lineopt_reconfiguration.py

@@ -291,6 +291,328 @@ def evolutionary_component_assignment(pcb_data, component_data, machine_number,
     return min(pop_val), population[np.argmin(pop_val)]
 
 
+class SpiderMonkeyOpt:
+    def __init__(self, pop_size, pcb_data, component_data, machine_number, estimator):
+        self.PcbData = pcb_data
+        self.ComponentData = component_data
+        self.Estimator = estimator
+
+        self.PopSize = pop_size
+        self.LocalLimit = pop_size
+        self.GlobalLimit = pop_size
+
+        self.MachineNum = machine_number
+        self.GroupSize = 0
+        # self.Dim = sum(data.fdn for _, data in component_data.iterrows()) + machine_number
+
+        self.CpPoints = defaultdict(int)
+        self.CpIndex = defaultdict(int)
+        self.CpNozzle = defaultdict(str)
+
+        self.Dim = 0
+        for cp_idx, data in component_data.iterrows():
+            # cp_feeders[cp_idx] = data.fdn
+
+            division_data = copy.deepcopy(data)
+            feeder_limit, total_points = division_data.fdn, division_data.points
+            surplus_points = total_points % feeder_limit
+            for _ in range(feeder_limit):
+                division_data.fdn, division_data.points = 1, math.floor(total_points / feeder_limit)
+                if surplus_points:
+                    division_data.points += 1
+                    surplus_points -= 1
+
+                self.CpPoints[self.Dim], self.CpNozzle[self.Dim] = division_data.points, division_data.nz
+                self.CpIndex[self.Dim] = cp_idx
+                self.Dim += 1
+
+        self.Dim += machine_number
+
+        component_list = list(range(len(self.CpPoints)))
+        self.GenPosition = []
+        self.GenPopVal = []
+        for _ in range(self.PopSize):
+            random.shuffle(component_list)
+            self.GenPosition.append(component_list.copy())
+            idx, prev = 0, 0
+            div = random.sample(range(len(component_list)), self.MachineNum - 1)
+            div.append(len(component_list))
+            div.sort(reverse=False)
+
+            for _ in range(1, self.MachineNum + 1):
+                self.GenPosition[-1].append(div[idx] - prev)
+                prev = div[idx]
+                idx += 1
+
+            self.GenPopVal.append(self.CalIndividualVal(self.GenPosition[-1]))
+
+        self.GroupPoint = [[0 for _ in range(2)] for _ in range(self.PopSize)]
+        self.GroupPart = 1
+
+        self.GenProb = None
+        self.GlobalMin = self.GenPopVal[0]
+        self.GlobalLeaderPosition = self.GenPosition[0].copy()
+        self.GlobalLimitCount = 0
+
+        self.LocalMin = np.ones(self.PopSize) * 1e10
+        self.LocalLeaderPosition = [[0 for _ in range(self.Dim)] for _ in range(self.PopSize)]
+        self.LocalLimitCount = [0 for _ in range(self.PopSize)]
+
+        for k in range(self.GroupSize):
+            self.LocalMin[k] = self.GenPopVal[int(self.GroupPoint[k, 0])]
+            self.LocalLeaderPosition[k,:] = self.GenPosition[int(self.GroupPoint[k, 0]),:]
+
+        self.CrossoverRatio = 0.1
+
+
+    def GlobalLearning(self):
+        GlobalTrial = self.GlobalMin
+        for i in range(self.PopSize):
+            if self.GenPopVal[i] < self.GlobalMin:
+                self.GlobalMin = self.GenPopVal[i]
+                self.GlobalLeaderPosition = self.GenPosition[i].copy()
+
+        if math.fabs(GlobalTrial - self.GlobalMin) < 1e-5:
+            self.GlobalLimitCount = self.GlobalLimitCount + 1
+        else:
+            self.GlobalLimitCount = 0
+
+    def LocalLearning(self):
+        OldMin = np.zeros(self.PopSize)
+        for k in range(self.GroupSize):
+            OldMin[k] = self.LocalMin[k]
+
+        for k in range(self.GroupSize):
+            i = int(self.GroupPoint[k][0])
+            while i <= int(self.GroupPoint[k][1]):
+                if self.GenPopVal[i] < self.LocalMin[k]:
+                    self.LocalMin[k] = self.GenPopVal[i]
+                    self.LocalLeaderPosition[k] = self.GenPosition[i].copy()
+                i = i + 1
+
+        for k in range(self.GroupSize):
+            if math.fabs(OldMin[k] - self.LocalMin[k]) < 1e-5:
+                self.LocalLimitCount[k] = self.LocalLimitCount[k] + 1
+            else:
+                self.LocalLimitCount[k] = 0
+
+    def CalculateProbabilities(self):
+        self.GenProb = [0 for _ in range(self.PopSize)]
+        MaxVal = self.GenPopVal[0]
+        i = 1
+        while i < self.PopSize:
+            if self.GenPopVal[i] > MaxVal:
+                MaxVal = self.GenPopVal[i]
+            i += 1
+        for i in range(self.PopSize):
+            self.GenProb[i] = (0.9 * (self.GenPopVal[i] / MaxVal)) + 0.1
+
+    def LocalLeaderPhase(self, k):
+        lo = int(self.GroupPoint[k][0])
+        hi = int(self.GroupPoint[k][1])
+        i = lo
+        while i <= hi:
+
+            NewGene1, NewGene2 = self.CrossoverOperation(self.GenPosition[i], self.LocalLeaderPosition[k])
+            NewGeneVal1, NewGeneVal2 = self.CalIndividualVal(NewGene1), self.CalIndividualVal(NewGene2)
+
+            if NewGeneVal1 < self.GenPopVal[i]:
+                self.GenPosition[i] = NewGene1
+                self.GenPopVal[i] = NewGeneVal1
+
+            if NewGeneVal2 < self.GenPopVal[i]:
+                self.GenPosition[i] = NewGene2
+                self.GenPopVal[i] = NewGeneVal2
+            i += 1
+
+    def GlobalLeaderPhase(self, k):
+        lo = int(self.GroupPoint[k][0])
+        hi = int(self.GroupPoint[k][1])
+        i = lo
+        l = lo
+        while l < hi:
+            if random.random() < self.GenProb[i]:
+                l += 1
+                NewGene1, NewGene2 = self.CrossoverOperation(self.GenPosition[i], self.GlobalLeaderPosition)
+                NewGeneVal1, NewGeneVal2 = self.CalIndividualVal(NewGene1), self.CalIndividualVal(NewGene2)
+
+                if NewGeneVal1 < self.GenPopVal[i]:
+                    self.GenPosition[i] = NewGene1
+                    self.GenPopVal[i] = NewGeneVal1
+
+                if NewGeneVal2 < self.GenPopVal[i]:
+                    self.GenPosition[i] = NewGene2
+                    self.GenPopVal[i] = NewGeneVal2
+
+            i += 1
+            if i == hi:
+                i = lo
+
+    def LocalLeaderDecision(self):
+        for k in range(self.GroupSize):
+            if self.LocalLimitCount[k] > self.LocalLimit:
+                i = self.GroupPoint[k][0]
+                while i <= int(self.GroupPoint[k][1]):
+                    if random.random() >= self.CrossoverRatio:
+                        NewGenPosition = list(range(self.Dim - self.MachineNum))
+                        random.shuffle(NewGenPosition)
+
+                        idx, prev = 0, 0
+                        div = random.sample(range(len(NewGenPosition)), self.MachineNum - 1)
+                        div.append(len(NewGenPosition))
+                        div.sort(reverse=False)
+
+                        for _ in range(1, self.MachineNum + 1):
+                            NewGenPosition.append(div[idx] - prev)
+                            prev = div[idx]
+                            idx += 1
+
+                        NewGenVal = self.CalIndividualVal(NewGenPosition)
+
+                        if NewGenVal < self.GenPopVal[i]:
+                            self.GenPosition[i] = NewGenPosition.copy()
+                            self.GenPopVal[i] = NewGenVal
+                    else:
+                        NewGene1, NewGene2 = self.CrossoverOperation(self.GenPosition[i], self.GlobalLeaderPosition)
+                        NewGeneVal1, NewGeneVal2 = self.CalIndividualVal(NewGene1), self.CalIndividualVal(NewGene2)
+
+                        if NewGeneVal1 < self.GenPopVal[i]:
+                            self.GenPosition[i] = NewGene1.copy()
+                            self.GenPopVal[i] = NewGeneVal1
+
+                        if NewGeneVal2 < self.GenPopVal[i]:
+                            self.GenPosition[i] = NewGene2.copy()
+                            self.GenPopVal[i] = NewGeneVal2
+
+                    i += 1
+
+                self.LocalLimitCount[k] = 0
+
+    def GlobalLeaderDecision(self):
+        if self.GlobalLimitCount> self.GlobalLimit:
+            self.GroupPart += 1
+            self.GlobalLimitCount = 0
+            self.CreateGroup()
+            self.LocalLearning()
+
+    def CreateGroup(self):
+        g = 0
+        lo = 0
+
+        while lo < self.PopSize:
+            hi = lo + int(self.PopSize / self.GroupPart)
+            self.GroupPoint[g][0] = lo
+            self.GroupPoint[g][1] = hi
+            if self.PopSize - hi < int(self.PopSize / self.GroupPart):
+                self.GroupPoint[g][1] = (self.PopSize - 1)
+            g = g + 1
+            lo = hi + 1
+        self.GroupSize = g
+
+    def CrossoverOperation(self, gene1, gene2):
+        len_ = len(gene1)
+        sub1, sub2 = partially_mapped_crossover(gene1[0: len_ - self.MachineNum], gene2[0: len_ - self.MachineNum])
+
+        pos1, pos2 = random.randint(0, self.MachineNum - 1), random.randint(0, self.MachineNum - 1)
+        machine_assign1, machine_assign2 = gene1[len_ - self.MachineNum:], gene2[len_ - self.MachineNum:]
+        machine_assign1[pos1], machine_assign2[pos2] = machine_assign2[pos2], machine_assign1[pos1]
+        while sum(machine_assign1) != len_ - self.MachineNum:
+            machine_idx = random.randint(0, self.MachineNum - 1)
+            if machine_assign1[machine_idx] == 0:
+                continue
+            if sum(machine_assign1) > len_ - self.MachineNum:
+                machine_assign1[machine_idx] -= 1
+            else:
+                machine_assign1[machine_idx] += 1
+
+        while sum(machine_assign2) != len_ - self.MachineNum:
+            machine_idx = random.randint(0, self.MachineNum - 1)
+            if machine_assign2[machine_idx] == 0:
+                continue
+            if sum(machine_assign2) > len_ - self.MachineNum:
+                machine_assign2[machine_idx] -= 1
+            else:
+                machine_assign2[machine_idx] += 1
+
+        sub1.extend(machine_assign1)
+        sub2.extend(machine_assign2)
+        return sub1, sub2
+
+
+    def CalIndividualVal(self, gene):
+        ComponentNum = len(self.ComponentData)
+        assignment_result = [[0 for _ in range(ComponentNum)] for _ in range(self.MachineNum)]
+
+        idx, machine_index = 0, 0
+        for num in range(self.Dim - self.MachineNum, self.Dim):
+            for _ in range(gene[num]):
+                assignment_result[machine_index][self.CpIndex[gene[idx]]] += self.CpPoints[gene[idx]]
+                idx += 1
+            machine_index += 1
+
+        val = 0
+        cp_items = converter(self.PcbData, self.ComponentData, assignment_result)
+        for machine_index in range(self.MachineNum):
+            cp_points, cp_nozzle, board_width, board_height = cp_items[machine_index]
+            val = max(val, self.Estimator.predict(cp_points, cp_nozzle, board_width, board_height))
+        return val
+
+def spider_monkey_component_assignment(pcb_data, component_data, machine_number, estimator):
+    population_size, iteration_number = 20, 50
+
+    smo = SpiderMonkeyOpt(population_size, pcb_data, component_data, machine_number, estimator)
+
+    # ========================== Calling: GlobalLearning() ======================== #
+    smo.GlobalLearning()
+
+    # ========================== Calling: create_group() ========================== #
+    smo.CreateGroup()
+
+    # ========================= Calling: LocalLearning() ========================== #
+    smo.LocalLearning()
+
+    # ================================= Looping ================================== #
+    with tqdm(total=iteration_number) as pbar:
+        pbar.set_description('spider monkey algorithm process for PCB assembly line balance')
+        for _ in range(iteration_number):
+            for k in range(smo.GroupSize):
+                # ==================== Calling: LocalLeaderPhase() =================== #
+                smo.LocalLeaderPhase(k)
+
+            # =================== Calling: CalculateProbabilities() ================== #
+            smo.CalculateProbabilities()
+
+            for k in range(smo.GroupSize):
+                # ==================== Calling: GlobalLeaderPhase() ================== #
+                smo.GlobalLeaderPhase(k)
+
+            # ======================= Calling: GlobalLearning() ====================== #
+            smo.GlobalLearning()
+
+            # ======================= Calling: LocalLearning() ======================= #
+            smo.LocalLearning()
+
+            # ================== Calling: LocalLeaderDecision() ====================== #
+            smo.LocalLeaderDecision()
+
+            # ===================== Calling: GlobalLeaderDecision() ================== #
+            smo.GlobalLeaderDecision()
+
+            pbar.update(1)
+
+    assignment_result = [[0 for _ in range(len(component_data))] for _ in range(machine_number)]
+
+    idx, machine_index = 0, 0
+    for num in range(smo.Dim - machine_number, smo.Dim):
+        for _ in range(smo.GlobalLeaderPosition[num]):
+            assignment_result[machine_index][smo.CpIndex[smo.GlobalLeaderPosition[idx]]] += \
+                smo.CpPoints[smo.GlobalLeaderPosition[idx]]
+            idx += 1
+        machine_index += 1
+
+    return smo.GlobalMin, assignment_result
+
+
 @timer_wrapper
 def line_optimizer_reconfiguration(component_data, pcb_data, machine_number):
     # === assignment of heads to modules is omitted ===
@@ -303,9 +625,8 @@ def line_optimizer_reconfiguration(component_data, pcb_data, machine_number):
             print('random component allocation algorithm process for PCB assembly line balance')
             val, assignment = random_component_assignment(pcb_data, component_data, machine_number, estimator)
         elif i == 1:
-            # brute force
-            # which is proved to be useless, since it only ran in reasonable time for the smaller test instances
-            continue
+            # spider monkey
+            val, assignment = spider_monkey_component_assignment(pcb_data, component_data, machine_number, estimator)
         elif i == 2:
             # local search
             print('local search component allocation algorithm process for PCB assembly line balance')
@@ -314,7 +635,8 @@ def line_optimizer_reconfiguration(component_data, pcb_data, machine_number):
             # evolutionary
             val, assignment = evolutionary_component_assignment(pcb_data, component_data, machine_number, estimator)
         else:
-            # greedy: unclear description
+            # brute force
+            # which is proved to be useless, since it only ran in reasonable time for the smaller test instances
             continue
 
         if optimal_val is None or val < optimal_val:
@@ -324,3 +646,5 @@ def line_optimizer_reconfiguration(component_data, pcb_data, machine_number):
         raise Exception('no feasible solution! ')
 
     return optimal_assignment
+
+

optimizer.py

@@ -1,7 +1,3 @@
-import time
-
-import pandas as pd
-
 from dataloader import *
 from lineopt_genetic import line_optimizer_genetic
 from lineopt_heuristic import line_optimizer_heuristic
@@ -12,9 +8,9 @@ from lineopt_model import line_optimizer_model
 from base_optimizer.optimizer_interface import *
 
 
-def optimizer(pcb_data, component_data, feeder_data, params):
+def optimizer(pcb_data, component_data, feeder_data, params, hinter=True):
     if params.machine_number == 1:
-        assembly_info = [base_optimizer(1, pcb_data, component_data, feeder_data, params, hinter=True)]
+        assembly_info = [base_optimizer(1, pcb_data, component_data, feeder_data, params, hinter=hinter)]
         return assembly_info
 
     if params.line_optimizer == 'hyper-heuristic' or params.line_optimizer == 'heuristic' or params.line_optimizer \
@@ -33,7 +29,7 @@ def optimizer(pcb_data, component_data, feeder_data, params):
         for machine_index in range(params.machine_number):
             assembly_info.append(base_optimizer(machine_index + 1, partial_pcb_data[machine_index],
                                                 partial_component_data[machine_index],
-                                                pd.DataFrame(columns=['slot', 'part']), params, hinter=True))
+                                                pd.DataFrame(columns=['slot', 'part']), params, hinter=hinter))
     elif params.line_optimizer == 'mip-model':
         assembly_info = line_optimizer_model(component_data, pcb_data, params.machine_number)
     else:
@@ -49,16 +45,14 @@ def main():
     parser.add_argument('--mode', default=1, type=int, help='mode: 0 -directly load pcb data without optimization '
                                                             'for data analysis, 1 -optimize pcb data, 2 -batch test')
     parser.add_argument('--filename', default='PCB.txt', type=str, help='load pcb data')
-    # parser.add_argument('--filename', default='chapter3-2/PCB2-8 Arg1.txt', type=str, help='load pcb data')
+
     parser.add_argument('--comp_register', default=1, type=int, help='register the component according the pcb data')
-    parser.add_argument('--machine_number', default=1, type=int, help='the number of machine in the assembly line')
-    # parser.add_argument('--machine_optimizer', default='mip-model', type=str, help='optimizer for single machine')
+    parser.add_argument('--machine_number', default=3, type=int, help='the number of machine in the assembly line')
     parser.add_argument('--machine_optimizer', default='feeder-priority', type=str, help='optimizer for single machine')
     parser.add_argument('--line_optimizer', default='hyper-heuristic', type=str, help='optimizer for PCB assembly line')
-    # parser.add_argument('--line_optimizer', default='model', type=str, help='optimizer for PCB assembly line')
     parser.add_argument('--feeder_limit', default=1, type=int, help='the upper feeder limit for each type of component')
     parser.add_argument('--save', default=0, type=int, help='save the optimization result')
-    parser.add_argument('--save_suffix', default='(10)', type=str, help='load pcb data')
+    parser.add_argument('--save_suffix', default='(1)', type=str, help='load pcb data')
     params = parser.parse_args()
 
     # 结果输出显示所有行和列
@@ -69,6 +63,7 @@ def main():
         assembly_info = []
         for machine_index in range(len(partial_pcb_data)):
             opt_res = convert_pcbdata_to_result(partial_pcb_data[machine_index], partial_component_data[machine_index])
+
             info = placement_info_evaluation(partial_component_data[machine_index], partial_pcb_data[machine_index],
                                              opt_res)
             assembly_info.append(info)
@@ -76,6 +71,7 @@ def main():
                                        nozzle_hinter=True, component_hinter=True, feeder_hinter=True)
 
             info.print()
+
             if params.save:
                 output_optimize_result(f'result/{params.filename[:-4]}-T-Solution-M0{machine_index + 1}',
                                        partial_component_data[machine_index], partial_pcb_data[machine_index], opt_res)
@@ -89,7 +85,6 @@ def main():
               f'standard deviation: {np.std([info.total_time for info in assembly_info]): .3f}')
 
     elif params.mode == 1:
-        # sys.stdout = open(f'record/{params.filename[:-4]}-{params.line_optimizer}.txt', 'w')
 
         # 加载PCB数据
         partial_pcb_data, partial_component_data, feeder_data = load_data(params.filename, load_feeder=True)
@@ -108,7 +103,10 @@ def main():
 
         print(f'finial assembly time: {max(info.total_time for info in assembly_info): .3f} s, '
               f'standard deviation: {np.std([info.total_time for info in assembly_info]): .3f}')
+
     elif params.mode == 2:
+
+        # sys.stdout = open(f'record/dissertation-experiment.txt', 'w')
         machine_optimizer = ['two-phase', 'hybrid-genetic', 'cell-division', 'feeder-priority', 'aggregation']
         running_round = 10
         opt_columns = ['Cycle', 'Pick', 'Nozzle-Change', 'Running-Time']
@@ -120,16 +118,18 @@ def main():
         for _, file in enumerate(os.listdir('data/')):
             if file[-3:] != 'txt':
                 continue
+
             partial_pcb_data, partial_component_data, feeder_data = load_data(file)
             pcb_data, component_data = merge_data(partial_pcb_data, partial_component_data)
+
             for opt in machine_optimizer:
                 for round_idx in range(running_round):
-                    print(f'--- file ：  {file},  round  ：  {round_idx},  optimizer ：  {opt} --- ')
+                    print(f'--- file ：  {file},  round  ：  {round_idx + 1},  optimizer ：  {opt} --- ')
 
-                    params = parser.parse_args(['--machine_optimizer', opt, '--machine_number', str(1)])
+                    params = parser.parse_args(['--machine_optimizer', opt, '--machine_number', str(1), '--filename', file])
 
                     start_time = time.time()
-                    assembly_info = optimizer(pcb_data, component_data, feeder_data, params)
+                    assembly_info = optimizer(pcb_data, component_data, feeder_data, params, hinter=False)
 
                     opt_result[opt].loc[file + str(round_idx + 1), 'Cycle'] = assembly_info[0].cycle_counter
                     opt_result[opt].loc[file + str(round_idx + 1), 'Pick'] = assembly_info[0].pickup_counter
@@ -140,8 +140,7 @@ def main():
             for opt, result in opt_result.items():
                 result.to_excel(writer, sheet_name=opt, float_format='%.3f', na_rep='')
     else:
-        # line_optimizer = ['T-Solution', 'hyper-heuristic', 'genetic', 'reconfiguration']
-        line_optimizer = ['genetic']
+        line_optimizer = ['T-Solution', 'hyper-heuristic', 'genetic', 'reconfiguration']
         file_dirs = ['L01', 'L02', 'L03']
 
         running_round = 10
@@ -169,6 +168,7 @@ def main():
                     warning_info = f'file: {file_dir}/{file}: an unexpected error occurs for data loader'
                     warnings.warn(warning_info, SyntaxWarning)
                     continue
+
                 machine_number = len(partial_pcb_data)
                 if not os.path.exists(f'record/{file_dir}'):
                     os.makedirs(f'record/{file_dir}')
@@ -207,11 +207,12 @@ def main():
                                 ['--filename', file_dir + '/' + file, '--machine_number', str(machine_number),
                                  '--line_optimizer', line_opt, '--save_suffix', f'({round_idx + 1})'])
                             start_time = time.time()
-                            assembly_info = optimizer(merge_pcb_data, merge_component_data, params)
+                            assembly_info = optimizer(merge_pcb_data, merge_component_data, None, params)
 
                             line_opt_result[file_dir].loc[file, line_opt + str(round_idx + 1)] = max(
                                 info.total_time for info in assembly_info)
                             line_opt_runtime[file_dir].loc[file, line_opt + str(round_idx + 1)] = time.time() - start_time
+
                             for machine_idx, info in enumerate(assembly_info):
                                 print(
                                     f'assembly time for machine {machine_idx + 1: d}: {info.total_time: .3f} s, '