update

Environment.py

@@ -84,6 +84,7 @@ class FMSEnv(ap.Model):
         if self.t >= self.int_stop_time:
             self.running = False
             self.stop()
+
     #     else:
     #
     # # print(f"running the {self.t} step")
@@ -121,22 +122,9 @@ def GA_run(inventory_bound=None):
     S = tuple(tuple([i, j]) for i, j in zip(material, inventory_bound[
                                                       len(pd.read_excel("initial_material.xlsx").to_numpy()): len(
                                                           pd.read_excel("initial_material.xlsx").to_numpy()) * 2]))
-    # print(s)
+
-    # print(S)
     dct_para = {
         'time': 300,  # 进行总时间数
-        # 'xv_int_max_order': random.randint(30, 50),
-        # 'xv_dlv_product_para': tuple([(30, 100), (30, 50)]),
-        # 'xv_dlv_product_para': tuple([30,40,30,20]), # 读取生产率 np.read.
-        # 'xv_int_dlv_period_lam': 8.5,
-        # 'xv_int_create_order_lam': 2,
-        # 'xv_ary_price_product': tuple([0.3,0.2,0.5,1]),
-        # 'xv_ary_cost_material_per': tuple([0.1,0.1,0.2,0.4]),
-        # 'xv_ary_volume_material': tuple([1.0, 1.5]),
-        # 'xv_ary_volume_product': tuple([3.0, 5.0]),
-        # 'xv_array_lead_time': 2,  # 读取原材料表格 np.read, 暂时不读 变量代表的含义
-        # 'xv_int_lead_time_c': 3,
-        # 'xv_int_lead_time_d': 1,
         'xv_ary_product_id': tuple(pd.read_excel("initial_product.xlsx").iloc[:, 0]),  # 产成品id顺序
         'xv_ary_material_id': tuple(pd.read_excel("initial_material.xlsx").iloc[:, 0]),  # 原材料id顺序
         'xv_product_num': len(pd.read_excel("initial_product.xlsx").to_numpy()),  # 产成品个数
@@ -150,43 +138,8 @@ def GA_run(inventory_bound=None):
         'xv_ary_s': s,  # s
         'xv_ary_S': S,  # S
         # 应读取遗传算法中随机生成的s，暂写为'1' 创建两个excel分别存储产品和原材料的库存 每个excel中存系统代码和库存
-        # 'xv_flt_initial_cash': 50000.0,
-        # 'dct_status_info': json.dumps({   #需要引入生产状态表
-        #     "0": {"xv_flt_produce_rate": tuple([0.0, 0.0]),
-        #           "xv_ary_mat_material": tuple([0.0, 0.0]),
-        #           "xv_flt_broken_rate": 0,
-        #           "xv_flt_run_cost": 0.0,
-        #           "name": "wait"
-        #           },
-        #     "1": {"xv_flt_produce_rate": tuple([90.0, 0.0]),
-        #           "xv_ary_mat_material": tuple([4.0, 1.0]),
-        #           "xv_flt_broken_rate": 0.03,
-        #           "xv_flt_run_cost": 40.0,
-        #           "name": "produceA"
-        #           },
-        #     "2": {"xv_flt_produce_rate": tuple([0.0, 60.0]),
-        #           "xv_ary_mat_material": tuple([1.5, 5.0]),
-        #           "xv_flt_broken_rate": 0.05,
-        #           "xv_flt_run_cost": 50.0,
-        #           "name": "produceB"
-        #           },
-        #     "3": {"xv_flt_produce_rate": tuple([55.0, 30.0]),
-        #           "xv_ary_mat_material": tuple([2.0, 1.5]),
-        #           "xv_flt_broken_rate": 0.07,
-        #           "xv_flt_run_cost": 60.0,
-        #           "name": "produceAB"
-        #           },
-        #     "-1": {"xv_flt_produce_rate": 0.0,
-        #            "xv_ary_mat_material": tuple([0.0, 0.0]),
-        #            "xv_flt_broken_rate": 0.1,
-        #            "xv_flt_run_cost": 100.0,
-        #            "name": "failed"
-        #            }
-        # })
-
     }
     sample = ap.Sample(dct_para)
-
     exp = ap.Experiment(FMSEnv, sample, iterations=1, record=True)
     results = exp.run()
     return results['variables']['FMSEnv']['op_os_all_delay_time'][dct_para['time']] / 2

ga_new.py

@@ -42,7 +42,7 @@ class GeneticAlgorithm:
     :ivar function: Object that can be used to evaluate the objective function
     """
 
-    def __init__(self, function, dim, lb, ub, int_var=None, pop_size=20, num_gen=300, start="Random"):
+    def __init__(self, function, dim, lb, ub, int_var=None, pop_size=6, num_gen=300, start="Random"):
 
         self.nvariables = dim  # column
         self.nindividuals = pop_size + (pop_size % 2)  # Make sure this is even row
@@ -102,6 +102,11 @@ class GeneticAlgorithm:
                 ind = np.where(population[:, i] > self.upper_boundary[i])
                 population[ind, i] -= 1
 
+        for pop in population:
+            for i in range(len(pop) // 2):
+                if pop[i] >= pop[i + len(pop) // 2]:
+                    pop[i], pop[i + len(pop) // 2] = pop[i + len(pop) // 2], pop[i]
+
         #  Evaluate all individuals
         # function_values = self.function(population) we cannot compute in this way to ensure x is one-dim in policy
         n_row, n_dim = population.shape
@@ -125,7 +130,8 @@ class GeneticAlgorithm:
         for _ in range(self.ngenerations):
             print('------------------------------')
             print("当前为第{}代".format(_))
-            print("最优个体为:{}".format(best_individual))
+            print("最优s为:{}".format(best_individual[0:115]))
+            print("最优S为:{}".format(best_individual[115:230]))
             print("最优值为:{}".format(best_value))
             print("------------------------------")
             # Do tournament selection to select the parents
@@ -174,6 +180,11 @@ class GeneticAlgorithm:
             # Keep the best individual
             population[0, :] = best_individual
 
+            for pop in population:
+                for i in range(len(pop) // 2):
+                    if pop[i] >= pop[i + len(pop) // 2]:
+                        pop[i], pop[i + len(pop) // 2] = pop[i + len(pop) // 2], pop[i]
+
             #  Evaluate all individuals
             # function_values = self.function(population) we cannot compute in this way to ensure x is one-dim in policy
             n_row, n_dim = population.shape