M3Opy/simLake.py at main · Critical-Infrastructure-Systems-Lab/M3Opy · GitHub

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
 # from main_script import sys_param
import sim
import numpy as np
import numpy.matlib

def simLake( q, h_in, policy, sys_param ):
    # global sys_param

    # % Simulation setting
    q_sim = np.append(np.nan, q )
    H = len(q_sim) - 1
    # % Initialization
    h = np.nan*np.ones(len(q_sim))
    s = np.nan*np.ones(len(q_sim))
    r = np.nan*np.ones(len(q_sim))
    u = np.nan*np.ones(len(q_sim))

    # % Start simulation
    h[0] = h_in
    # s[0] = sim.levelToStorage(h[0],sys_param)

    A  = sys_param['simulation']['A']
    h0 = sys_param['simulation']['h0']
    s[0] = A*(h[0] - h0)

    for t in range(H):
# =============================================================================
      # % Compute release decision

        if sys_param['algorithm']['name'] == 'rand':
          r_min = sys_param['simulation']['r_min']
          r_max = sys_param['simulation']['r_max']

          uu = r_min + rand*(r_max - r_min)

# =============================================
        elif sys_param['algorithm']['name'] == 'doe':
          uf = rand;
          r_min = sys_param['simulation']['r_min'];
          r_max = sys_param['simulation']['r_max'];
          w = sys_param['simulation']['w'];

          if uf < 0.2:
            uu = r_min + rand*(r_max*0.66 - r_min);
          elif uf < 0.40:
            uu = r_max*0.66 + rand*(r_max - r_max*0.66);
          else:
            uu = w

# =============================================
        elif sys_param['algorithm']['name'] ==  'ddp':
          import ddp
          discr_s = sys_param['algorithm']['discr_s'];
          discr_q = sys_param['algorithm']['discr_q'];
          discr_u = sys_param['algorithm']['discr_u'];

          min_rel = sys_param['algorithm']['min_rel'];
          max_rel = sys_param['algorithm']['max_rel'];
          # print('simLake: min rel=',min_rel,'max rel=',max_rel)
          w = sys_param['simulation']['w'];

          idx_q  = np.argmin( np.absolute( discr_q - q_sim[t+1] ) );

          # % Minimum and maximum release for current storage and inflow:
          sys_param['simulation']['vv'] = np.interp( s[t], discr_s , min_rel[: , idx_q] )
          sys_param['simulation']['VV'] = np.interp( s[t], discr_s , max_rel[: , idx_q] )

          _ , idx_u  = ddp.Bellman_ddp( policy[:,t+1] , s[t] , q_sim[t+1] , sys_param)
          # % Choose one decision value (idx_u can return multiple equivalent decisions)
          uu = sim.extractor_ref( idx_u , discr_u , w )

# =============================================
        elif sys_param['algorithm']['name'] ==  'sdp':
          import sdp
          discr_s = sys_param['algorithm']['discr_s'];
          discr_q = sys_param['algorithm']['discr_q'];
          discr_u = sys_param['algorithm']['discr_u'];

          min_rel = sys_param['algorithm']['min_rel'];
          max_rel = sys_param['algorithm']['max_rel'];

          w = sys_param['simulation']['w'];

          # % Minimum and maximum release for current storage and inflow:
          idx_q  = np.argmin( np.absolute( discr_q - q_sim[t+1] ) )

          v = np.interp( s[t], discr_s , min_rel[: , idx_q] )
          sys_param['simulation']['vv'] = np.matlib.repmat( v, 1, len(discr_q) ).flatten()

          V = np.interp( s[t], discr_s , max_rel[: , idx_q] )
          sys_param['simulation']['VV'] = np.matlib.repmat( V, 1, len(discr_q) ).flatten()
          # print(sys_param['simulation']['VV'].size)
          # print(sys_param['simulation']['VV'])
          # print('s[t]=',s[t])

          _ , idx_u  = sdp.Bellman_sdp( policy , s[t] , sys_param )
          # print('idx_u in simlake', idx_u)

          # % Choose one decision value (idx_u can return multiple equivalent
          # % decisions)
          uu = sim.extractor_ref( idx_u , discr_u , w )
          # print('uu1=',uu, type(uu))

  # =============================================
        elif sys_param['algorithm']['name'] ==  'emodps':
          import emodps

          policy_class = sys_param['algorithm']['policy_class']

          if policy_class == 'stdOP':
             uu = emodps.std_operating_policy(h[t], policy, sys_param)
          else:
             raise Exception('Policy class not defined.\
            Please check or modify this function to use a different\
            class of parameterized functions')

  # =============================================
        elif sys_param['algorithm']['name'] ==  'fqi':
          discr_s = sys_param['algorithm']['discr_s'];
          discr_u = sys_param['algorithm']['discr_u'];

          w = sys_param['simulation']['w'];

          _, idx_u = readQ(s[t], discr_s, discr_u, policy['Q']);
          uu = sim.extractor_ref( idx_u , discr_u , w );

# =============================================
        elif sys_param['algorithm']['name'] ==  'ssdp':
          interp_foo = sys_param['algorithm']['interp_foo'];
          T          = sys_param['algorithm']['T'];
          discr_u    = sys_param['algorithm']['discr_u'];
          discr_s    = sys_param['algorithm']['discr_s'];
          discr_q    = sys_param['algorithm']['discr_q'];
          min_rel    = sys_param['algorithm']['min_rel'];
          max_rel    = sys_param['algorithm']['max_rel'];
          esp_sample = sys_param['algorithm']['esp_sample'];

          t_idx = mod[t-1, T] + 1

          w = sys_param['simulation']['w']

          _, idx_q  = np.minimum(np.absolute( discr_q - q_sim[t+1] ))

          sys_param['simulation']['vv'] = interp_foo( discr_s, min_rel [: , idx_q] , s[t] )
          sys_param['simulation']['VV'] = interp_foo( discr_s, max_rel [: , idx_q] , s[t] )

          # idx_u = reopt_ssdp( policy.H(:,:,t_idx+1), s[t], esp_sample(t_idx,:) );
#           uu = sim.extractor_ref( idx_u , discr_u, w );

# =============================================
        elif sys_param['algorithm']['name'] ==  'iso' :
          regressorName = sys_param['algorithm']['regressorName']

          if regressorName == 'linear_spline':
             uu = policy.predict(h[t])
          else:
             raise Exception('Regressor not defined. Please check or modify this \
                        function to use a different regression method')

# =============================================
        else:
          uu = np.nan

        u[t] = uu
        # print('u=',u)
       # % Hourly integration of mass-balance equation
        # print('s[0]=',s[0])
        s[t+1], r[t+1] = sim.massBalance( s[t], u[t], q_sim[t+1], sys_param )
        # print('s=',s)
        # h[t+1] = sim.storageToLevel(s[t+1],sys_param)
        h[t+1] = s[t+1]/A + h0


# # =============================================================================
    # % Calculate objectives (daily average of immediate costs)
    # print(s)
    # return
    g_flo, g_irr = sim.immediate_costs(h[1:], r[1:], sys_param);

    Jflo = np.mean(g_flo)
    Jirr = np.mean(g_irr)
    # print('h', h)

    sys_param['algorithm']['h'] = h
    sys_param['algorithm']['s'] = s
    sys_param['algorithm']['r'] = r
    sys_param['algorithm']['u'] = u

    # print(policy)
    # return

    return Jflo, Jirr, h, u#, r, g_flo, g_irr