PPO.py

@@ -331,8 +331,8 @@ if __name__ == '__main__':
         buffer = list()
         '''init training parameters'''
         num_episode = args.num_episode
-        # args.train = False
-        # args.save_network = False
+        args.train = False
+        args.save_network = False
         # args.test_network = False
         # args.save_test_data = False
         # args.compare_with_gurobi = False

environment.py

@@ -1,7 +1,7 @@
 import gym
 import numpy as np
 import pandas as pd
-
+from gym import spaces
 from module import *
 from parameters import *
 from data_manager import *
@@ -39,8 +39,8 @@ class ESSEnv(gym.Env):
         self.solar = Solar(self.solar_parameters)
         self.wind = Wind(self.wind_parameters)
 
-        self.action_space = gym.spaces.Box(low=-1, high=1, shape=(5,), dtype=np.float32)  # 已增加调节电压动作
-        self.state_space = gym.spaces.Box(low=0, high=1, shape=(10,), dtype=np.float32)
+        self.action_space = spaces.Box(low=-1, high=1, shape=(5,), dtype=np.float32)  # 已增加调节电压动作
+        self.state_space = spaces.Box(low=-np.inf, high=np.inf, shape=(10,), dtype=np.float32)
 
     def reset(self, *args):
         self.month = np.random.randint(1, 13)  # choose 12 month
@@ -69,7 +69,7 @@ class ESSEnv(gym.Env):
         temperature = self.data_manager.get_temperature_data(self.month, self.day, self.current_time)
         irradiance = self.data_manager.get_irradiance_data(self.month, self.day, self.current_time)
         wind_speed = self.data_manager.get_wind_data(self.month, self.day, self.current_time)
-
+        # print('house_load:', house_load)
         pv_generation = self.solar.step(temperature, irradiance)
         wd_generation = self.wind.step(wind_speed)
         generation = pv_generation + wd_generation
@@ -129,9 +129,10 @@ class ESSEnv(gym.Env):
         solar_cost = self.solar.get_cost(self.solar.current_power)
         wind_cost = self.wind.gen_cost(self.wind.current_power)
 
+        reward -= (battery_cost + dg1_cost + dg2_cost + dg3_cost + solar_cost + wind_cost + excess_penalty +
+                   deficient_penalty - sell_benefit + buy_cost) / 1e3
         self.operation_cost = (battery_cost + dg1_cost + dg2_cost + dg3_cost + solar_cost + wind_cost + excess_penalty +
                                deficient_penalty - sell_benefit + buy_cost)
-        reward -= self.operation_cost / 1e3
         self.unbalance = unbalance
         self.real_unbalance = self.shedding + self.excess
         final_step_outputs = [self.dg1.current_output, self.dg2.current_output, self.dg3.current_output,

plotDRL.py

@@ -133,8 +133,8 @@ def plot_evaluation_information(datasource, directory):
     axs[1, 1].cla()
     axs[1, 1].set_ylabel('Costs')
     axs[1, 1].bar(eval_data['time_step'], eval_data['operation_cost'])
-    fig.savefig(f"{directory}/evaluation_information.svg", format='svg', dpi=600, bbox_inches='tight')
-    print('evaluation figure have been ploted and saved')
+    fig.savefig(f"{directory}/Evoluation Information.svg", format='svg', dpi=600, bbox_inches='tight')
+    print('evaluation figure have been plot and saved')
 
 
 def make_dir(directory, feature_change):

tools.py

@@ -43,8 +43,6 @@ def optimization_base_result(env, month, day, initial_soc):
     NUM_GEN = len(DG_parameters.keys())
     battery_capacity = env.battery.capacity
     battery_efficiency = env.battery.efficiency
-    solar_cofficient = env.solar.opex_cofficient
-    wind_cofficient = env.wind.opex_cofficient
 
     m = gp.Model("UC")
 
@@ -84,10 +82,8 @@ def optimization_base_result(env, month, day, initial_soc):
         t in range(period) for g in range(NUM_GEN))
     cost_grid_import = gp.quicksum(grid_energy_import[t] * price[t] for t in range(period))
     cost_grid_export = gp.quicksum(grid_energy_export[t] * price[t] * env.sell_coefficient for t in range(period))
-    cost_solar = gp.quicksum(pv[t] * solar_cofficient for t in range(period))
-    cost_wind = gp.quicksum(wind[t] * wind_cofficient for t in range(period))
 
-    m.setObjective((cost_gen + cost_grid_import - cost_grid_export + cost_solar + cost_wind), GRB.MINIMIZE)
+    m.setObjective((cost_gen + cost_grid_import - cost_grid_export), GRB.MINIMIZE)
     m.optimize()
 
     output_record = {'pv': [], 'wind': [], 'price': [], 'load': [], 'netload': [],
@@ -99,13 +95,11 @@ def optimization_base_result(env, month, day, initial_soc):
                        for g in range(NUM_GEN))
         grid_import_cost = grid_energy_import[t].x * price[t]
         grid_export_cost = grid_energy_export[t].x * price[t] * env.sell_coefficient
-        solar_cost = pv[t] * solar_cofficient
-        wind_cost = wind[t] * wind_cofficient
         output_record['pv'].append(pv[t])
         output_record['wind'].append(wind[t])
         output_record['price'].append(price[t])
         output_record['load'].append(load[t])
-        output_record['netload'].append(load[t] - pv[t] - wind[t])
+        output_record['netload'].append(load[t] - pv[t])
         output_record['soc'].append(soc[t].x)
         output_record['battery_energy_change'].append(battery_energy_change[t].x)
         output_record['grid_import'].append(grid_energy_import[t].x)
@@ -113,7 +107,7 @@ def optimization_base_result(env, month, day, initial_soc):
         output_record['gen1'].append(gen_output[0, t].x)
         output_record['gen2'].append(gen_output[1, t].x)
         output_record['gen3'].append(gen_output[2, t].x)
-        output_record['step_cost'].append(gen_cost + grid_import_cost - grid_export_cost + solar_cost + wind_cost)
+        output_record['step_cost'].append(gen_cost + grid_import_cost - grid_export_cost)
     output_record_df = pd.DataFrame.from_dict(output_record)
     return output_record_df