多维时序 | MATLAB实现TPA-LSTM(时间注意力注意力机制长短期记忆神经网络)多输入单输出_matlab lstm 多维变量输入

多维时序 | MATLAB实现TPA-LSTM(时间注意力注意力机制长短期记忆神经网络)多输入单输出

预测效果

基本介绍

注意力机制模仿人脑，更加注重重要信息，而忽略相对无用的信息，已被广泛应用于自然语言处理、图像及语音识别中，近年来也被广泛应用于各类预测问题。传统注意力机制注重不同时间点的权重分布，在每个时间步只含有一个变量时有较好的效果。但对于区域内的多风电机组功率预测，每个时间步都含有多个变量，各个变量之间可能存在复杂的非线性内在联系，且每个变量序列都有自己的特征和周期，难以单独选取某个时间步作为注意重点。而TPA则由多个一维CNN滤波器从BiLSTM隐藏状态行向量抽取特征，使得模型能够从不同时间步学习多变量之间的互相依赖关系。

环境介绍

运行环境，Matlab2020b。

程序设计

• 完整程序下载：TPA-LSTM

% 数据集 列为特征，行为样本数目
%% 数据导入及处理
load('./Train.mat')
Train.weekend = dummyvar(Train.weekend);
Train.month = dummyvar(Train.month);
Train = movevars(Train,{'weekend','month'},'After','demandLag');
Train.ts = [];

% Train.hour = dummyvar(Train.hour);
%自己主动观察右侧工作区变量格式，对前面数据进行更改替换
Train(1,:) =[];
y = Train.demand;
x = Train{:,2:5};
[xnorm,xopt] = mapminmax(x',0,1);
[ynorm,yopt] = mapminmax(y',0,1);
%
% xnorm = [xnorm;Train.weekend';Train.month'];
%%
% x = x';
xnorm = xnorm(:,1:1000);
ynorm = ynorm(1:1000);

k = 24;           % 滞后长度

% 转换成2-D image
for i = 1:length(ynorm)-k

    Train_xNorm(:,i,:) = xnorm(:,i:i+k-1);
    Train_yNorm(i) = ynorm(i+k-1);
    Train_y(i) = y(i+k-1);
end
Train_yNorm= Train_yNorm';



ytest = Train.demand(1001:1170);
xtest = Train{1001:1170,2:5};
[xtestnorm] = mapminmax('apply', xtest',xopt);
[ytestnorm] = mapminmax('apply',ytest',yopt);
% xtestnorm = [xtestnorm; Train.weekend(1001:1170,:)'; Train.month(1001:1170,:)'];
xtest = xtest';
for i = 1:length(ytestnorm)-k
    Test_xNorm(:,i,:) = xtestnorm(:,i:i+k-1);
    Test_yNorm(i) = ytestnorm(i+k-1);
    Test_y(i) = ytest(i+k-1);
end
Test_yNorm = Test_yNorm';

clear k i x y
%
Train_xNorm = dlarray(Train_xNorm,'CBT');
Train_yNorm = dlarray(Train_yNorm,'BC');
Test_xNorm = dlarray(Test_xNorm,'CBT');
Test_yNorm = dlarray(Test_yNorm,'BC');
%% 训练集和验证集划分
TrainSampleLength = length(Train_yNorm);
validatasize = floor(TrainSampleLength * 0.1);
Validata_xNorm = Train_xNorm(:,end - validatasize:end,:);
Validata_yNorm = Train_yNorm(:,TrainSampleLength-validatasize:end);
Validata_y = Train_y(TrainSampleLength-validatasize:end);

%参数设置
inputSize = size(Train_xNorm,1);   %数据输入x的特征维度
outputSize = 1;                    %数据输出y的维度  
numhidden_units1=50;

[params,~] = paramsInit(numhidden_units1,inputSize,outputSize);     % 导入初始化参数

[~,validatastate] = paramsInit(numhidden_units1,inputSize,outputSize);     % 导入初始化参数
[~,TestState] = paramsInit(numhidden_units1,inputSize,outputSize);     % 导入初始化参数
% 训练相关参数
TrainOptions;
numIterationsPerEpoch = floor((TrainSampleLength-validatasize)/minibatchsize);
LearnRate = 0.01;
%% Loop over epochs.
figure
start = tic;
lineLossTrain = animatedline('color','r');
validationLoss = animatedline('color',[0 0 0]./255,'Marker','o','MarkerFaceColor',[150 150 150]./255);
xlabel("Iteration")
ylabel("Loss")

% epoch 更新 
iteration = 0;
for epoch = 1 : numEpochs
   
    [~,state] = paramsInit(numhidden_units1,inputSize,outputSize);       % 每轮epoch，state初始化
    disp(['Epoch: ', int2str(epoch)])  
      % 作图（训练过程损失图）--------------------------********————————————————————————————————————————————————
        D = duration(0,0,toc(start),'Format','hh:mm:ss');
        addpoints(lineLossTrain,iteration,double(gather(extractdata(loss))))
        if iteration == 1 || mod(iteration,validationFrequency) == 0
            addpoints(validationLoss,iteration,double(gather(extractdata(lossValidation))))
        end
        title("Epoch: " + epoch + ", Elapsed: " + string(D))
        legend('训练集','验证集')
        drawnow
        
    end
    
    % 每轮epoch 更新学习率
    if mod(epoch,5) == 0
        LearnRate = LearnRate * LearnRateDropFactor;
    end
end


%% 训练集
Predict_yNorm = TPAModelPredict(gpuArray(Train_xNorm),params,TestState);
Predict_yNorm = extractdata(Predict_yNorm);

Predict_y = mapminmax('reverse',Predict_yNorm,yopt);
%
figure
plot(Predict_y,'-.','Color',[50 100 180]./255,'linewidth',1.5,'Markersize',3,'MarkerFaceColor',[50 100 180]./255);
hold on 
plot(Train_y,'-.','Color',[150 150 150]./255,'linewidth',1.5,'Markersize',3,'MarkerFaceColor',[150 150 150]./255)
legend('训练集预测值','训练集实际值')
%% 验证集
Predict_yNorm = TPAModelPredict(gpuArray(Validata_xNorm),params,TestState);
Predict_yNorm = extractdata(Predict_yNorm);

Predict_y = mapminmax('reverse',Predict_yNorm,yopt);
%
figure
plot(Predict_y,'-.','Color',[255 0 0]./255,'linewidth',1.5,'Markersize',3,'MarkerFaceColor',[255 0 0]./255);
hold on 
plot(Validata_y,'--','Color',[150 150 150]./255,'linewidth',1.5,'Markersize',3,'MarkerFaceColor',[0 0 0]./255)
legend('验证集预测值','验证集实际值')

%%  predict(传统递归测试集)
% clear Predict_yNorm
% % Test_xNorm = extractdata(Test_xNorm);
% for i = 1:size(Test_xNorm,2)
%     if i ==1
%         [a,Teststate] = TPAModelPredict(dlarray(Test_xNorm(:,i,:),'CBT'),params,TestState);

%% predict(直接测试集)
Predict_yNorm = TPAModelPredict(dlarray(Test_xNorm,'CBT'),params,TestState);
Predict_YNorm = extractdata(Predict_yNorm);
Predict_y = mapminmax('reverse',Predict_YNorm,yopt);

figure
plot(Predict_y,'-.','Color',[0 0 255]./255,'linewidth',1.5,'Markersize',3,'MarkerFaceColor',[0 0 255]./255);
hold on 
plot(Test_y,'--','Color',[150 150 150]./255,'linewidth',1.5,'Markersize',3,'MarkerFaceColor',[0 0 0]./255)
legend('测试集预测值','测试集实际值')
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

参考资料

[1] https://blog.csdn.net/kjm13182345320/article/details/125644313?spm=1001.2014.3001.5501
[2] https://blog.csdn.net/kjm13182345320/article/details/125637228?spm=1001.2014.3001.5501
[3] https://download.csdn.net/download/kjm13182345320/85661169?spm=1001.2014.3001.5501