def calTime(filename):
"""
使用Q表来计算完成所有任务消耗的时间
:param filename:
:return:
"""
task = createTask()
env = Maze(task)
RL = QLearningTable(actions=list(range(env.n_actions)), filename=filename)
Time1 = []
# Time2 = []
for i in range(10000):
observation = env.reset()
while True:
action = RL.choose_action_real(str(observation))
observation_, reward, done = env.step(action)
# print(observation,action,reward)
observation = observation_
if done:
time1 = findmax(task)
# time2 = calOmegaT(task,np.array([255])[0])
Time1.append(time1)
# Time2.append(time2)
break
# print(np.mean(Time1))
# print(np.mean(Time2))
return np.mean(Time1)
E1 = E
t += Baseline(user, G, E1, KE, edgeuser)
random.append(t / 10)
# 穷举法
H = topology() # 任务图
E2 = E
t = Brute(user, H, E2, KE, edgeuser) # (用户，应用，能量，能量限制)
brute.append(t)
# Q-learning
observation = env.reset()
stade = copy.deepcopy(observation)
while True:
# RL 大脑根据 state 的观测值挑选 action
action = RL.choose_action_real(str(observation))
# 探索者在环境中实施这个 action, 并得到环境返回的下一个 state 观测值, reward 和 done (是否是掉下地狱或者升上天堂)
observation_, reward, done = env.step(action, user)
# RL 从这个序列 (state, action, reward, state_) 中学习
RL.learn(str(stade), action, reward, str(observation_))
# 将下一个 state 的值传到下一次循环
stade = copy.deepcopy(observation_)
# 如果掉下地狱或者升上天堂, 这回合就结束了
if done:
break
q_learning.append(7000 - reward)
print(reward)
x = [5, 10, 15, 20, 25, 30, 35, 40, 45]
plt.plot(x, random, marker='o', label="$Random$", c='b')
plt.plot(x, brute, marker='x', label="$Brute-force$", c='y')
