along/HPCC2025
1
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

修改离散环境，连续不动给惩罚

Browse Source
This commit is contained in:
weixin_46229132 2025-03-19 20:58:34 +08:00
parent 4972306ca7
commit 3dba6e4a53
2 changed files with 18 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Duel_Double_DQN/utils.py
View File

@ -14,7 +14,7 @@ def evaluate_policy(env, agent, turns = 3):
action_series.append(a) action_series.append(a)
total_scores += r total_scores += r
s = s_next s = s_next
print('action series: ', np.roudn(action_series, 3)) print('action series: ', np.round(action_series, 3))
print('state: ', s) print('state: ', s)
return int(total_scores/turns) return int(total_scores/turns)

26
env_dis.py
View File

@ -209,10 +209,11 @@ class PartitionMazeEnv(gym.Env):
elif self.phase == 2: elif self.phase == 2:
# 阶段 2：路径规划（走迷宫） # 阶段 2：路径规划（走迷宫）
reward = 0
# 后 4 个动作对应上下左右移动 # 后 4 个动作对应上下左右移动
current_car = self.current_car_index current_car = self.current_car_index
current_row, current_col = self.reverse_rectangles[self.car_pos[current_car]] current_row, current_col = self.reverse_rectangles[self.car_pos[current_car]]
# 初始化新的行、列为当前值 # 初始化新的行、列为当前值
new_row, new_col = current_row, current_col new_row, new_col = current_row, current_col
@ -227,21 +228,28 @@ class PartitionMazeEnv(gym.Env):
else: else:
# 无效动作，保持原地 # 无效动作，保持原地
pass pass
# 检查是否移动
car_moved = (new_row != current_row or new_col != current_col)
# 更新车辆位置 # 更新车辆位置
self.car_pos[current_car] = self.rectangles[( self.car_pos[current_car] = self.rectangles[(
new_row, new_col)]['center'] new_row, new_col)]['center']
if new_row != current_row or new_col != current_col: if car_moved:
self.car_traj[current_car].append((new_row, new_col)) self.car_traj[current_car].append((new_row, new_col))
self.step_count += 1
self.current_car_index = (
self.current_car_index + 1) % self.num_cars
# 更新访问标记：将新网格标记为已访问 # 更新访问标记：将新网格标记为已访问
self.rectangles[(new_row, new_col)]['is_visited'] = True self.rectangles[(new_row, new_col)]['is_visited'] = True
# 观察状态 # 记录所有车辆一轮中是否移动
reward = 0 if self.current_car_index == 0:
# 如果是新一轮的开始，初始化移动标记
self.cars_moved = [False] * self.num_cars
self.cars_moved[current_car] = car_moved
# 如果一轮结束，检查是否所有车辆都没有移动
if self.current_car_index == (self.num_cars - 1) and not any(self.cars_moved):
reward -= 10 # 扣除 10 分奖励
self.step_count += 1
self.current_car_index = (
self.current_car_index + 1) % self.num_cars
max_regions = (self.CUT_NUM // 2 + 1) ** 2 max_regions = (self.CUT_NUM // 2 + 1) ** 2
visit_status = np.zeros(max_regions, dtype=np.float32) visit_status = np.zeros(max_regions, dtype=np.float32)