-
大小: 5KB文件類(lèi)型: .rar金幣: 2下載: 2 次發(fā)布日期: 2021-06-18
- 語(yǔ)言: Html/CSS
- 標(biāo)簽: MatLab??神經(jīng)網(wǎng)絡(luò)??機(jī)器學(xué)習(xí)??倒立擺??強(qiáng)化學(xué)習(xí)??
資源簡(jiǎn)介
MatLab強(qiáng)化學(xué)習(xí)代碼包,使用深度Q學(xué)習(xí)(神經(jīng)網(wǎng)絡(luò))控制倒立擺的代碼。 詳細(xì)說(shuō)明可參看我的專(zhuān)欄《強(qiáng)化學(xué)習(xí)與控制》 https://blog.csdn.net/weixin_43723517/category_9676083.html "I thought what I'd do was I'd pretend I was one of those deaf-mutes, or should I?"

代碼片段和文件信息
%輸出控制力矩列表
FcTable=[-3-2.5-2-1.5-1-0.5-0.300.30.511.522.53];
global?N_Fc
N_Fc=size(FcTable);?N_Fc=N_Fc(2);
%DQN神經(jīng)網(wǎng)絡(luò)初始化
QNet_eval=fitnet([4040]);
%訓(xùn)練神經(jīng)網(wǎng)絡(luò),規(guī)定輸入輸出
S_ini=150;
Iniset=zeros(6S_ini);??%前五行為輸入,最后一行為目標(biāo)輸出。第一行X,第二行dotX,第三行θ,第四行dotθ,第五行Fc
Iniset(1:)=0.5*rand(1S_ini)-0.5*rand(1S_ini);
Iniset(2:)=0.1*rand(1S_ini)-0.1*rand(1S_ini);
Iniset(3:)=(2*rand(1S_ini)-2*rand(1S_ini))*pi/180;????????????%單位為弧度
Iniset(4:)=(2*rand(1S_ini)-2*rand(1S_ini))*pi/180;????????????%單位為弧度
Iniset(5:)=unidrnd(N_Fc1S_ini);
for?i=1:S_ini
????Iniset(6i)=Reward_Cal(Iniset(1:4i));
end?
QNet_eval=train(QNet_evalIniset(1:5:)Iniset(6:));
QNet_target=QNet_eval;
%將神經(jīng)網(wǎng)絡(luò)訓(xùn)練算法設(shè)置為自適應(yīng)動(dòng)量梯度下降法
QNet_eval.trainFcn=‘traingdx‘;??
%關(guān)閉訓(xùn)練圖窗nntraintool
%QNet_eval.trainParam.showWindow=0;
%學(xué)習(xí)參數(shù)初始化
global?alpha?gamma
alpha=0.55;?????????????????%學(xué)習(xí)率alpha
gamma=0.99;?????????????????%獎(jiǎng)勵(lì)傳遞效率gamma
nBatch=400;?????????????????%神經(jīng)網(wǎng)絡(luò)單次訓(xùn)練調(diào)用的樣本量
N_gap=400;??????????????????%神經(jīng)網(wǎng)絡(luò)訓(xùn)練的頻率
N_renew=3*N_gap;????????????%目標(biāo)神經(jīng)網(wǎng)絡(luò)更新的頻率
%Replaymemory初始化???
S_memo=4000;????????????????%MemoSize
Rmemo=zeros(9S_memo);??????%1-4:CPstate5act6-9:newstate
Memopointer=1;??????????????%Replay?memory的寫(xiě)入指針
%}
?屬性????????????大小?????日期????時(shí)間???名稱(chēng)
-----------?---------??----------?-----??----
?????文件???????1403??2020-03-04?13:22??CartPoleDQN_DEMO\AgentInitializer.m
?????文件???????1155??2020-03-04?10:14??CartPoleDQN_DEMO\CalculationQtarget.m
?????文件????????280??2020-03-04?13:24??CartPoleDQN_DEMO\CartPoleInitializer.m
?????文件????????377??2020-03-04?09:18??CartPoleDQN_DEMO\CartPoleReset.m
?????文件???????3291??2020-03-07?16:44??CartPoleDQN_DEMO\CartPole_DQN.m
?????文件????????619??2020-03-03?10:35??CartPoleDQN_DEMO\CartPole_Eqs.m
?????文件????????157??2020-02-25?11:40??CartPoleDQN_DEMO\PointerMove.m
?????文件????????204??2020-03-04?13:50??CartPoleDQN_DEMO\Reward_Cal.m
?????文件????????422??2020-03-04?10:28??CartPoleDQN_DEMO\tcegreedy.m
?????目錄??????????0??2020-03-07?14:57??CartPoleDQN_DEMO
-----------?---------??----------?-----??----
?????????????????7908????????????????????10
評(píng)論
共有 條評(píng)論