基于长短期神经网络的异常值检测，基于LSTM的异常值检测（代码完整，数据齐全）

clc clear close all rng(2) [a,ax,ay] = xlsread('数据 .xlsx'); a(:,7:end) = []; [idx,c,sumd ,d] = kmeans(a, 2); pan2 = sum(idx==2); pan1 = sum(idx==1); num = [a idx]; y = zeros(size(a)); figure for ii = 1:6 x = a(:,ii); y(:,ii) = hampel(x,10); subplot(6,1,ii) plot(x,'k-') hold on plot(y(:,ii),'r-') if ii == 1 title('异常值检测') end end dx = a(:,1)-y(:,1); [mx,my] = find(dx>20); for ii = length(mx):-1:1 num(mx(ii),:) = []; end n = randperm(length(num)); m=12000; input_train =num(n(1:m),1:6 )';%训练数据输出数 output_train = num(n(1:m),7)';%训练数据输入数据 input_test = num((12001:end),1:6)';%测试数据输出数据 output_test = num((12001:end),7)';%测试数据输入数据 [inputn,inputps]=mapminmax(input_train,-1,1);%训练数据的输入数据的归一化 [outputn,outputps]=mapminmax(output_train,-1,1);%训练数据的输出数据的归一化de inputn_test=mapminmax('apply',input_test,inputps); %% Define Network Architecture % Define the network architecture. numFeatures = 6;%输入层维度 numResponses = 1;%输出维度 % 200 hidden units numHiddenUnits = 125;%第一层维度 % a fully connected layer of size 50 & a dropout layer with dropout probability 0.5 layers = [ ... sequenceInputLayer(numFeatures)%输入层 lstmLayer(numHiddenUnits,'OutputMode','sequence')%第一层 fullyConnectedLayer(110)%链接层 dropoutLayer(0.1)%遗忘层 fullyConnectedLayer(numResponses)%链接层 regressionLayer];%回归层 % Specify the training options. % Train for 60 epochs with mini-batches of size 20 using the solver 'adam' maxEpochs = 60;%最大迭代次数 miniBatchSize =100;%最小批量 % the learning rate == 0.01 % set the gradient threshold to 1 % set 'Shuffle' to 'never'never options = trainingOptions('adam', ... %解算器 'MaxEpochs',maxEpochs, ... %最大迭代次数 'MiniBatchSize',miniBatchSize, ... %最小批次 'InitialLearnRate',0.01, ... %初始学习率 'GradientThreshold',inf, ... %梯度阈值 'Shuffle','every-epoch', ... %打乱顺序 'Plots','training-progress',... %画图 'Verbose',0); %不输出训练过程 %% Train the Network net = trainNetwork(inputn,outputn,layers,options);%开始训练 % save maynet.mat %% Test the Network y_pred = predict(net,inputn_test,'MiniBatchSize',miniBatchSize)';%测试仿真输出 [y_pred]=(mapminmax('reverse',y_pred',outputps)); % y_pred0 = predict(net,inputn,'MiniBatchSize',1)';%训练拟合值 % y_pred0=(mapminmax('reverse',y_pred0',outputps))'; y_pred=round(double(y_pred)); zhunquelv = sum(y_pred==output_test)/length(output_test) error1 = y_pred-output_test;%误差 % zz=(mapminmax('reverse',zz,inputps)); figure%打开一个图像窗口 plot(output_test,'ro')%红色实线，点的形状为o hold on plot(y_pred,'b*')%黑色实线，点的形状为* hold off%停止画图测试图')%Y轴名称 legend('实际值','预测输出')%标签 ylabel('类别') set(gca,'fontsize',12) set(gca,'YTick',1:2) set(gca,'YTickLabel',{'运行正常','运行故障'}) figure plot(error1,'k-*') title('测试误差图') R_2 = R_2(output_test,y_pred) [MSE,RMSE,MBE,MAE ] =MSE_RMSE_MBE_MAE(output_test,y_pred); result_table = table; result_table.simd = y_pred'; result_table.trued = output_test'; writetable(result_table,'./结果2.csv')