-
Notifications
You must be signed in to change notification settings - Fork 10
/
Copy pathHybrid1.m
143 lines (118 loc) · 4.37 KB
/
Hybrid1.m
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
clear;
clc;
%%
% GLOBAL PARAMETERS
% Parameter values
num_episodes = 1024;
numValidationExperiments = 20;
%%
% Buck Boost Converter Parameters
V_source_value = 48;
L_inductance = 10e-6;
C_capacitance = 40e-3;
R_load = 100;
%%
gain_k = 100;
integral_I = 350000;
period_val = 0.00001;
pw_percent = 50;
%%
% Signal Processing Parameters
prev_time = 0;
init_action = 1;
stopping_criterion = 1000;
threshold1= 0.4;
threshold2 =1;
error_threshold = 0.02;
%%
Ts = 0.00001;
Tf = 0.3;
V_ref =110;%30;%80%110;
%%
% RL Parameters
miniBatch_percent = 0.8;
learnRateActor = 0.05;
learnRateCritic= 0.05;
criticLayerSizes= [256 256];
actorLayerSizes= [256 256];
discountFactor= 0.995;
max_steps = ceil(Tf/Ts);
ExperienceHorisonLength = 10;
ClipFactorVal = 0.2;
EntropyLossWeightVal = 0.05;
MiniBatchSizeVal = ceil(ExperienceHorisonLength*miniBatch_percent);
NumEpochsVal = 5;
DiscountFactorVal = 0.99;
%%
% RL Agent
mdl = 'DCDC_BBC_hybrid1';
open_system(mdl)
agentblk = [mdl '/RL Agent'];
numObs = 3; % [v0, e, de/dt]
observationInfo = rlNumericSpec([numObs,1],...
'LowerLimit',[-inf -inf 0]',...
'UpperLimit',[0.1 V_ref inf]');
observationInfo.Name = 'observations';
observationInfo.Description = 'integrated error, error, and measured height';
numObservations = observationInfo.Dimension(1);
a = [0;1];
actionInfo = rlFiniteSetSpec(a);
env = rlSimulinkEnv(mdl,agentblk,observationInfo,actionInfo);
env.ResetFcn = @(in) setVariable(in,'init_action',1);
num_inputs = numObs;
criticNetwork = [imageInputLayer([num_inputs 1 1],'Normalization','none','Name','state')
fullyConnectedLayer(criticLayerSizes(1),'Name','CriticFC1',...
'Weights',sqrt(2/numObs)*(rand(criticLayerSizes(1),numObs)-0.5), ...
'Bias',1e-3*ones(criticLayerSizes(1),1))
reluLayer('Name','CriticRelu1')
fullyConnectedLayer(criticLayerSizes(2),'Name','CriticFC2',...
'Weights',sqrt(2/criticLayerSizes(1))*(rand(criticLayerSizes(2),criticLayerSizes(1))-0.5), ...
'Bias',1e-3*ones(criticLayerSizes(2),1))
reluLayer('Name','CriticRelu2')
fullyConnectedLayer(1,'Name','CriticOutput',...
'Weights',sqrt(2/criticLayerSizes(2))*(rand(1,criticLayerSizes(2))-0.5), ...
'Bias',1e-3)];
criticOpts = rlRepresentationOptions('LearnRate',learnRateCritic,'GradientThreshold',1);
critic = rlValueRepresentation(criticNetwork,observationInfo,'Observation',{'state'},criticOpts);
numAct = numel(actionInfo.Elements);
actorNetwork = [imageInputLayer([numObs 1 1],'Normalization','none','Name','observation')
fullyConnectedLayer(actorLayerSizes(1),'Name','ActorFC1',...
'Weights',sqrt(2/numObs)*(rand(actorLayerSizes(1),numObs)-0.5), ...
'Bias',1e-3*ones(actorLayerSizes(1),1))
reluLayer('Name','ActorRelu1')
fullyConnectedLayer(actorLayerSizes(2),'Name','ActorFC2',...
'Weights',sqrt(2/actorLayerSizes(1))*(rand(actorLayerSizes(2),actorLayerSizes(1))-0.5), ...
'Bias',1e-3*ones(actorLayerSizes(2),1))
reluLayer('Name','ActorRelu2')
fullyConnectedLayer(numAct,'Name','Action',...
'Weights',sqrt(2/actorLayerSizes(2))*(rand(numAct,actorLayerSizes(2))-0.5), ...
'Bias',1e-3*ones(numAct,1))
softmaxLayer('Name','actionProbability')
];
actorOpts = rlRepresentationOptions('LearnRate',learnRateActor,'GradientThreshold',1);
actor = rlStochasticActorRepresentation(actorNetwork,observationInfo,actionInfo,...
'Observation',{'observation'},actorOpts);
agentOpts = rlPPOAgentOptions('ExperienceHorizon',ExperienceHorisonLength,...
'ClipFactor',ClipFactorVal,...
'EntropyLossWeight',0.02,...
'MiniBatchSize',MiniBatchSizeVal,...
'NumEpoch',NumEpochsVal,...
'AdvantageEstimateMethod','finite-horizon',...
'GAEFactor',0.98,...
'SampleTime',Ts,...
'DiscountFactor',DiscountFactorVal);
%%
agent = rlPPOAgent(actor,critic,agentOpts);
trainOpts = rlTrainingOptions(...
'MaxEpisodes',num_episodes,...
'MaxStepsPerEpisode',max_steps,...
'Verbose',true,...
'Plots','training-progress',...
'StopTrainingCriteria','AverageReward',...
'StopTrainingValue',inf,...
'ScoreAveragingWindowLength',50,...
'SaveAgentCriteria',"EpisodeReward",...
'SaveAgentValue',50000);
%%
% Train Agent
trainingStats = train(agent,env,trainOpts);