-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathtest_env_hvac.py
314 lines (230 loc) · 10 KB
/
test_env_hvac.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
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
import numpy as np
import pytest
import tensorflow as tf
from tests.conftest import sample_state, sample_action
def test_str(hvac):
print()
print(hvac)
def test_repr(hvac):
print()
print(repr(hvac))
def test_temp_bounds(hvac):
assert hvac.temp_lower_bound.shape == [hvac.state_size, 1]
assert hvac.temp_upper_bound.shape == [hvac.state_size, 1]
assert tf.reduce_all(hvac.temp_lower_bound <= hvac.temp_upper_bound)
def test_thermal_conductance(hvac):
assert hvac.R_outside.shape == [hvac.state_size, 1]
assert hvac.R_hall.shape == [hvac.state_size, 1]
assert hvac.R_wall.shape == [hvac.state_size, hvac.state_size]
assert tf.reduce_all(hvac.R_outside > 0.0)
assert tf.reduce_all(hvac.R_hall > 0.0)
assert tf.reduce_all(hvac.R_wall > 0.0)
def test_adj(hvac):
adj = hvac.adj
assert adj.shape == [hvac.state_size, hvac.state_size]
assert tf.reduce_all(tf.linalg.diag_part(adj) == False)
assert tf.reduce_all(tf.linalg.band_part(adj, -1, 0) == False)
def test_transition(hvac):
batch_size = 10
state1 = sample_state(hvac)
action1 = sample_action(hvac)
state2 = sample_state(hvac, batch_size=batch_size)
action2 = sample_action(hvac, batch_size=batch_size)
for state, action in [(state1, action1), (state2, action2)]:
next_state = hvac.transition(state, action)
assert next_state.shape == state.shape
fn = hvac.transition
cfn1 = fn.get_concrete_function(state1, action1)
cfn2 = fn.get_concrete_function(state2, action2)
assert cfn1 == cfn2
def test_linear_transition(hvac):
state1 = sample_state(hvac)
action1 = sample_action(hvac)
batch_size = 10
state2 = sample_state(hvac, batch_size=batch_size)
action2 = sample_action(hvac, batch_size=batch_size)
for state, action in [(state1, action1), (state2, action2)]:
model = hvac.get_linear_transition(state, action)
# check f
next_state = hvac.transition(state, action)
assert model.f.shape == next_state.shape
assert tf.reduce_all(model.f == next_state)
# check f_x
air = action * hvac.air_max
heating_x = - tf.linalg.diag(tf.squeeze(air * hvac.CAP_AIR, axis=-1))
adj = tf.logical_or(hvac.adj, tf.transpose(hvac.adj))
adj = tf.cast(adj, tf.float32)
conduction_between_rooms_x = - adj / hvac.R_wall * (-tf.ones_like(adj))
adj_outside = tf.cast(hvac.adj_outside, tf.float32)
conduction_with_outside_x = - tf.linalg.diag(
tf.squeeze(adj_outside / hvac.R_outside))
adj_hall = tf.cast(hvac.adj_hall, tf.float32)
conduction_with_hall_x = - tf.linalg.diag(
tf.squeeze(adj_hall / hvac.R_hall))
C = tf.linalg.diag(tf.squeeze(hvac.TIME_DELTA / hvac.capacity))
expected_f_x = (
tf.eye(hvac.state_size)
+ tf.matmul(C,
heating_x
+ conduction_between_rooms_x
+ conduction_with_outside_x
+ conduction_with_hall_x
)
)
# print(expected_f_x)
# print(model.f_x)
assert expected_f_x.shape == model.f_x.shape
# assert tf.reduce_all(tf.abs(model.f_x - expected_f_x) < 1e-3)
# check f_u
temp = state
expected_f_u = tf.linalg.diag(
tf.squeeze(
hvac.TIME_DELTA / hvac.capacity * hvac.air_max * hvac.CAP_AIR * (hvac.TEMP_AIR - temp),
axis=-1))
assert model.f_u.shape == expected_f_u.shape
assert tf.reduce_all(tf.abs(model.f_u - expected_f_u) < 1e-3)
fn = hvac.get_linear_transition
cfn1 = fn.get_concrete_function(state1, action1)
cfn2 = fn.get_concrete_function(state2, action2)
assert cfn1 == cfn2
def test_conduction_between_rooms(hvac):
state1 = sample_state(hvac)
state2 = sample_state(hvac, batch_size=10)
for state in [state1, state2]:
batch_size = tf.shape(state)[0]
state_size = tf.shape(state)[1]
conduction = hvac._conduction_between_rooms(state)
assert conduction.shape == state.shape
adj = tf.logical_or(hvac.adj, tf.transpose(hvac.adj))
R_wall = hvac.R_wall
temp = tf.squeeze(state, axis=-1)
expected = np.empty(shape=[batch_size, state_size], dtype=np.float32)
for i in range(batch_size):
for s in range(state_size):
c = 0.0
for p in range(state_size):
c += float(adj[s, p]) * (temp[i, p] - temp[i, s]) / R_wall[s, p]
expected[i, s] = c
expected = tf.constant(expected,
shape=[batch_size, state_size, 1],
dtype=tf.float32)
assert expected.shape == conduction.shape
assert tf.reduce_all(tf.abs(conduction - expected) < 1e-3)
fn = hvac._conduction_between_rooms
cfn1 = fn.get_concrete_function(state1)
cfn2 = fn.get_concrete_function(state2)
assert cfn1 == cfn2
def test_conduction_with_outside(hvac):
state1 = sample_state(hvac)
state2 = sample_state(hvac, batch_size=10)
for state in [state1, state2]:
batch_size = tf.shape(state)[0]
conduction = hvac._conduction_with_outside(state)
assert conduction.shape == state.shape
for i in range(batch_size):
conduction_ = conduction[i]
state_ = state[i]
idx1 = tf.where(tf.squeeze(hvac.adj_outside))
cond = tf.gather(tf.squeeze(conduction_), idx1)
temp_diff = tf.gather(tf.squeeze(hvac.temp_outside - state_), idx1)
assert tf.reduce_all(tf.sign(cond) == tf.sign(temp_diff))
idx2 = tf.where(tf.squeeze(tf.logical_not(hvac.adj_outside)))
cond = tf.gather(tf.squeeze(conduction_), idx2)
assert tf.reduce_all(cond == 0.0)
fn = hvac._conduction_with_outside
cfn1 = fn.get_concrete_function(state1)
cfn2 = fn.get_concrete_function(state2)
assert cfn1 == cfn2
def test_conduction_with_hall(hvac):
state1 = sample_state(hvac)
state2 = sample_state(hvac, batch_size=10)
for state in [state1, state2]:
batch_size = tf.shape(state)[0]
conduction = hvac._conduction_with_hall(state)
assert conduction.shape == state.shape
for i in range(batch_size):
conduction_ = conduction[i]
state_ = state[i]
idx1 = tf.where(tf.squeeze(hvac.adj_hall))
cond = tf.gather(tf.squeeze(conduction_), idx1)
temp_diff = tf.gather(tf.squeeze(hvac.temp_hall - state_), idx1)
assert tf.reduce_all(tf.sign(cond) == tf.sign(temp_diff))
idx2 = tf.where(tf.squeeze(tf.logical_not(hvac.adj_hall)))
cond = tf.gather(tf.squeeze(conduction_), idx2)
assert tf.reduce_all(cond == 0.0)
fn = hvac._conduction_with_hall
cfn1 = fn.get_concrete_function(state1)
cfn2 = fn.get_concrete_function(state2)
assert cfn1 == cfn2
def test_cost(hvac):
state1 = sample_state(hvac)
action1 = sample_action(hvac)
batch_size = 10
state2 = sample_state(hvac, batch_size=batch_size)
action2 = sample_action(hvac, batch_size=batch_size)
for state, action in [(state1, action1), (state2, action2)]:
batch_size = tf.shape(state)[0]
cost = hvac.cost(state, action)
assert cost.shape == [batch_size,]
assert tf.reduce_all(cost >= 0.0)
fn = hvac.cost
cfn1 = fn.get_concrete_function(state1, action1)
cfn2 = fn.get_concrete_function(state2, action2)
assert cfn1 == cfn2
def test_final_cost(hvac):
state1 = sample_state(hvac)[0]
state2 = sample_state(hvac)[0]
for state in [state1, state2]:
cost = hvac.final_cost(state)
assert cost.shape == []
assert cost >= 0.0
fn = hvac.final_cost
cfn1 = fn.get_concrete_function(state1)
cfn2 = fn.get_concrete_function(state2)
assert cfn1 == cfn2
def test_quadratic_cost(hvac):
state1 = sample_state(hvac)
action1 = sample_action(hvac)
batch_size = 10
state2 = sample_state(hvac, batch_size=batch_size)
action2 = sample_action(hvac, batch_size=batch_size)
for state, action in [(state1, action1), (state2, action2)]:
batch_size = tf.shape(state)[0]
model = hvac.get_quadratic_cost(state, action)
# check l
assert tf.reduce_all(model.l == hvac.cost(state, action))
# check l_x
temp = state
below_limit = - tf.linalg.diag(
tf.squeeze(tf.sign(tf.maximum(0.0, hvac.temp_lower_bound - temp)),
axis=-1))
above_limit = tf.linalg.diag(
tf.squeeze(tf.sign(tf.maximum(0.0, temp - hvac.temp_upper_bound)),
axis=-1))
out_of_bounds_penalty_x = (
hvac.PENALTY * (below_limit + above_limit)
)
set_point_penalty_x = - hvac.SET_POINT_PENALTY * tf.linalg.diag(
tf.squeeze(tf.sign((hvac.temp_lower_bound + hvac.temp_upper_bound) / 2 - temp),
axis=-1))
expected_l_x = tf.reduce_sum(
out_of_bounds_penalty_x + set_point_penalty_x,
axis=-1,
keepdims=True)
assert model.l_x.shape == expected_l_x.shape
assert tf.reduce_all(tf.abs(model.l_x - expected_l_x) < 1e-3)
# check l_u
expected_l_u = tf.reshape(
tf.tile(hvac.air_max * hvac.COST_AIR, [batch_size, 1]),
shape=[batch_size, -1, 1])
assert model.l_u.shape == expected_l_u.shape
assert tf.reduce_all(tf.abs(model.l_u - expected_l_u) < 1e-3)
# check all 2nd order derivatives are null
assert tf.reduce_all(model.l_xx == tf.zeros([hvac.state_size, hvac.state_size]))
assert tf.reduce_all(model.l_uu == tf.zeros([hvac.state_size, hvac.state_size]))
assert tf.reduce_all(model.l_xu == tf.zeros([hvac.state_size, hvac.state_size]))
assert tf.reduce_all(model.l_ux == tf.zeros([hvac.state_size, hvac.state_size]))
fn = hvac.get_quadratic_cost
cfn1 = fn.get_concrete_function(state1, action1)
cfn2 = fn.get_concrete_function(state2, action2)
assert cfn1 == cfn2