-
Notifications
You must be signed in to change notification settings - Fork 0
/
Env_PID.py
282 lines (241 loc) · 9.96 KB
/
Env_PID.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
# 输入各个库
from Model_5_18 import LSTM
from Model_width_5_18 import LSTM_Width
import numpy as np
import pyglet
import PID
import time
import matplotlib
matplotlib.use("TkAgg")
# 设置超参数
KP = 0.1
KI = 0.1
KD = 0.1
KP2 = 0.1
KI2 = 0.1
KD2 = 0.1
learning_rate = 1.0
height_lstm = LSTM()
height_lstm.restore()
width_lstm = LSTM_Width()
width_lstm.restore()
filename1 = "./Data/prediction_net_Height1.txt"
dataset1 = np.loadtxt(filename1)
filename2 = './Data/prediction_net_Width1.txt'
dataset2 = np.loadtxt(filename2)
dataset = np.zeros((8166,4))
dataset[:,0] = dataset1[:,0]
dataset[:,1] = dataset1[:,1]
dataset[:,2] = dataset1[:,2]
dataset[:,3] = dataset2[:,2]
dataset = np.array(dataset)
class Welding_Env():# 该函数的作用是搭建焊接环境，供强化学习训练和验证时交互使用
viewer = None
def __init__(self):# 初始化
## define the original values of the variables
self.Kp = KP
self.Ki = KI
self.Kd = KD
self.Kp2 = KP2
self.Ki2 = KI2
self.Kd2 = KD2
self.H_error = 0
self.W_error = 0
self.H_del_error = 0
self.W_del_error = 0
self.H_error_last = 0
self.W_error_last = 0
self.H_target = 0
self.W_target = 0
self.H_prediction = 0
self.W_prediction = 0
self.H_actual = 0
self.W_actual = 0
self.on_goal = 0
self.goal = 0
self.action_bound = [-1, 1]
self.action_dim = 6
self.state_dim = 8
self.counter = 0
# !!!!! wrong coding
self.Rs_list = [8] *200
self.Ws_list = [12] *200
self.Rs = 8
self.Ws = 12
self.delta_A = 0
self.delta_B = 0
self.delta_Rs = 0
self.delta_Ws = 0
self.flag = 0
def step(self, action):# 焊接过程走一步的函数，即根据强化学习的动作值得出下一步焊道的形状
done = False
self.counter += 1
# obtain error and delta error
self.H_error = self.H_target - self.H_prediction# 误差值
self.H_del_error = self.H_error - self.H_error_last# 误差值的差分
self.H_error_last = self.H_error
self.W_error = self.W_target - self.W_prediction# 误差值
self.W_del_error = self.W_error - self.W_error_last# 误差值的差分
self.W_error_last = self.W_error
# pid control the robot speed
# 底层为PID控制器，强化学习的作用为调节PID控制器的参数1
# pid1 = PID.PID(self.Kp + learning_rate * action[0]*0.1, self.Ki + learning_rate * action[1]*0.1, self.Kd + learning_rate * action[2]*0.1 )
# pid2 = PID.PID(self.Kp2 + learning_rate * action[3]*0.1, self.Ki2 + learning_rate * action[4]*0.1, self.Kd2 + learning_rate * action[5]*0.1)
pid1 = PID.PID(self.Kp, self.Ki, self.Kd)
pid2 = PID.PID(self.Kp2, self.Ki2 , self.Kd2)
action_np = np.zeros((6,1),dtype="float32")
action_np[0,0] = action[0]
action_np[1,0] = action[1]
action_np[2,0] = action[2]
action_np[3,0] = action[3]
action_np[4, 0] = action[4]
action_np[5, 0] = action[5]
self.delta_A = pid1.update(self.H_error,self.H_del_error)
self.delta_B = pid2.update(self.W_error,self.W_del_error)
# 28组数据反解
self.delta_Rs = -12.1674 * self.delta_A + 3.0336 * self.delta_B
self.delta_Ws = -26.0291 * self.delta_A + 2.1745 * self.delta_B
self.Rs += self.delta_Rs
self.Ws += self.delta_Ws
self.Rs = np.clip(self.Rs, 3, 13)
self.Ws = np.clip(self.Ws, self.Rs, 2*self.Rs)
print(self.Rs,self.Ws)
self.H_actual = height_lstm.welding_pred(self.Rs_list[-height_lstm.TIMESTEPS:],
self.Ws_list[-height_lstm.TIMESTEPS:])
self.W_actual = width_lstm.welding_pred(self.Rs_list[-width_lstm.TIMESTEPS:],
self.Ws_list[-width_lstm.TIMESTEPS:])
for num in range(0, 60):
self.Rs_list.append(self.Rs[0]) ## wrong ???
self.Ws_list.append(self.Ws[0])
# print("original:",np.shape(self.Rs_list))
# 将新的焊接参数序列输入焊接过程模型，得到新的焊道形状预测值
self.H_prediction = height_lstm.welding_pred(self.Rs_list[-height_lstm.TIMESTEPS:], self.Ws_list[-height_lstm.TIMESTEPS:])
self.W_prediction = width_lstm.welding_pred(self.Rs_list[-width_lstm.TIMESTEPS:], self.Ws_list[-width_lstm.TIMESTEPS:])
# print(self.Rs_list[-1 : ], self.Ws_list[-1 : ], self.H_prediction, self.target)
for num in range(0, 59):
self.Rs_list.pop() ## wrong ???
self.Ws_list.pop()
# print("after:", np.shape(self.Rs_list))
# reward
# if abs(self.H_target - self.H_prediction) < 0.002 and abs(self.W_target - self.W_prediction) < 0.005:
# self.on_goal += 1
# r = 1
# if self.on_goal > 80:
# done = True
# else:
# r = 1 / (1 + np.exp((abs(self.H_target - self.H_prediction) + abs(self.W_target - self.W_prediction)))) - 0.5
# # r = - (abs(self.H_target - self.H_prediction) + abs(self.W_target - self.W_prediction))
# # self.on_goal = 0
if abs(self.H_target - self.H_prediction) < 0.002 and abs(self.W_target - self.W_prediction) < 0.005:
self.on_goal += 1
r = 1
if self.on_goal > 80:
done = True
else:
r = 1 / (1 + np.exp(abs(self.H_target - self.H_prediction) + abs(self.W_target - self.W_prediction))) - 0.5
self.on_goal = 0
# if self.flag>60:
# done = True
# r = -100
# state
s = np.hstack(((self.H_prediction - 2), self.H_error, self.H_del_error, ((self.W_prediction - 5) / 3),self.W_error, self.W_del_error, (self.Rs - 8) / 5, ((self.Ws - 14.5) / 11.5)))
# s = np.hstack(((self.H_prediction - 2), self.H_target, ((self.W_prediction - 5) / 3),
# self.W_target))
#print(s,r)
return s, r, done
# set the initilize values
def reset(self):
self.flag = 0
self.H_error = 0
self.H_del_error = 0
self.H_error_last = 0
self.W_error = 0
self.W_del_error = 0
self.W_error_last = 0
self.Rs = 8
self.Ws = 12
#self.Wf = 10
# self.H_prediction = np.random.uniform(low=1.65, high=2.4, size=1)
# self.W_prediction = np.random.uniform(low=4.7, high=8, size=1)
# self.H_target = np.random.uniform(low=1.65, high=2.4, size=1)
# self.W_target = np.random.uniform(low=4.7, high=8, size=1)
ran_num1 = np.random.randint(0,8166)
self.H_prediction = [dataset[ran_num1,2]]
self.W_prediction = [dataset[ran_num1,3]]
ran_num2 = np.random.randint(0, 8166)
self.H_target =[dataset[ran_num2,2]]
self.W_target = [dataset[ran_num2,3]]
# self.H_prediction = [dataset[ran_num,2]]
# print("self.H:",self.H_prediction)
self.H_prediction = np.array(self.H_prediction)
self.W_prediction = np.array(self.W_prediction)
self.H_target = np.array(self.H_target)
self.W_target = np.array(self.W_target)
s = np.hstack(((self.H_prediction - 2), self.H_error, self.H_del_error, ((self.W_prediction - 5) / 3),self.W_error, self.W_del_error, (self.Rs - 8) / 5, ((self.Ws - 14.5) / 11.5)))
# s = np.hstack(((self.H_prediction - 2), self.H_target, ((self.W_prediction - 5) / 3),
# self.W_target))
return s
def render(self):
if self.viewer is None:
self.viewer = Viewer(self.H_prediction, self.H_target)
self.viewer.render(self.H_prediction, self.H_target)
def sample_action(self):
return np.random.rand(2) - 0.5
class Viewer(pyglet.window.Window):# 显示函数
def __init__(self, Y_t, goal):
# vsync=False to not use the monitor FPS, we can speed up training
super(Viewer, self).__init__(width=400, height=400, resizable=False, caption='Arm', vsync=False)
pyglet.gl.glClearColor(1, 1, 1, 1)
self.Y_t = Y_t
# print("init")
self.goal_info = goal
self.batch = pyglet.graphics.Batch() # display whole batch at once
self.goal = self.batch.add(
4, pyglet.gl.GL_QUADS, None, # 4 corners
('v2f', [100, 100 + self.goal_info, # location
100, 105 + self.goal_info,
300, 105 + self.goal_info,
300, 100 + self.goal_info]),
('c3B', (86, 109, 249) * 4)) # color
self.arm1 = self.batch.add(
4, pyglet.gl.GL_QUADS, None,
('v2f', [250, 250, # location
250, 255,
255, 255,
255, 250]),
('c3B', (249, 86, 86) * 4,)) # color
def render(self, H_prediction, target):
self.Y_t = H_prediction
self._update_arm(H_prediction, target)
self.switch_to()
self.dispatch_events()
self.dispatch_event('on_draw')
self.flip()
# print(self.goal_info['h'])
def on_draw(self):
self.clear()
self.batch.draw()
def _update_arm(self, H_prdiction, target):
# update goal
self.goal.vertices = (
100, 100 + target * 10,
100, 105 + target * 10,
300, 105 + target * 10,
300, 100 + target * 10)
# update arm
#height = self.Y_t
#print(H_prdiction)
self.arm1.vertices = (
195, 100 + H_prdiction * 10,
195, 105 + H_prdiction * 10,
205, 105 + H_prdiction * 10,
205, 100 + H_prdiction * 10)
if __name__ == '__main__':
env = Welding_Env()
while True:
print("new epoch")
s = env.reset()
for i in range(100):
env.render()
env.step(env.sample_action())
time.sleep(0.01)