基于卷积-双向长短期记忆网络结合SE注意力机制(CNN-BiLSTM-SE Attention)的回归预测Matlab完整程序

%% 清空环境变量 warning off % 关闭报警信息 close all % 关闭开启的图窗 clear % 清空变量 clc % 清空命令行 %% 导入数据 res = xlsread('数据集.xlsx'); %% 数据分析 num_size = 0.7; % 训练集占数据集比例 outdim = 1; % 最后一列为输出 num_samples = size(res, 1); % 样本个数 res = res(randperm(num_samples), :); % 打乱数据集（不希望打乱时，注释该行） num_train_s = round(num_size * num_samples); % 训练集样本个数 f_ = size(res, 2) - outdim; % 输入特征维度 %% 划分训练集和测试集 P_train = res(1: num_train_s, 1: f_)'; T_train = res(1: num_train_s, f_ + 1: end)'; M = size(P_train, 2); P_test = res(num_train_s + 1: end, 1: f_)'; T_test = res(num_train_s + 1: end, f_ + 1: end)'; N = size(P_test, 2); %% 数据归一化 [p_train, ps_input] = mapminmax(P_train, 0, 1); p_test = mapminmax('apply', P_test, ps_input); [t_train, ps_output] = mapminmax(T_train, 0, 1); t_test = mapminmax('apply', T_test, ps_output); %% 数据平铺 % 将数据平铺成1维数据只是一种处理方式 % 也可以平铺成2维数据，以及3维数据，需要修改对应模型结构 % 但是应该始终和输入层数据结构保持一致 p_train = double(reshape(p_train, f_, 1, 1, M)); p_test = double(reshape(p_test , f_, 1, 1, N)); t_train = double(t_train)'; t_test = double(t_test )'; %% 数据格式转换 for i = 1 : M Lp_train{i, 1} = p_train(:, :, 1, i); end for i = 1 : N Lp_test{i, 1} = p_test( :, :, 1, i); end %% 建立模型 lgraph = layerGraph(); % 建立空白网络结构 tempLayers = [ sequenceInputLayer([f_, 1, 1], "Name", "sequence") % 建立输入层，输入数据结构为[f_, 1, 1] sequenceFoldingLayer("Name", "seqfold")]; % 建立序列折叠层 lgraph = addLayers(lgraph, tempLayers); % 将上述网络结构加入空白结构中 tempLayers = convolution2dLayer([3, 1], 32, "Name", "conv_1"); % 卷积层 卷积核[3, 1] 步长[1, 1] 通道数 32 lgraph = addLayers(lgraph,tempLayers); % 将上述网络结构加入空白结构中 tempLayers = [ reluLayer("Name", "relu_1") % 激活层 convolution2dLayer([3, 1], 64, "Name", "conv_2") % 卷积层 卷积核[3, 1] 步长[1, 1] 通道数 64 reluLayer("Name", "relu_2")]; % 激活层 lgraph = addLayers(lgraph, tempLayers); % 将上述网络结构加入空白结构中 tempLayers = [ globalAveragePooling2dLayer("Name", "gapool") % 全局平均池化层 fullyConnectedLayer(16, "Name", "fc_2") % SE注意力机制，通道数的1 / 4 reluLayer("Name", "relu_3") % 激活层 fullyConnectedLayer(64, "Name", "fc_3") % SE注意力机制，数目和通道数相同 sigmoidLayer("Name", "sigmoid")]; % 激活层 lgraph = addLayers(lgraph, tempLayers); % 将上述网络结构加入空白结构中 tempLayers = multiplicationLayer(2, "Name", "multiplication"); % 点乘的注意力 lgraph = addLayers(lgraph, tempLayers); % 将上述网络结构加入空白结构中 tempLayers = [ sequenceUnfoldingLayer("Name", "sequnfold") % 建立序列反折叠层 flattenLayer("Name", "flatten") % 网络铺平层 bilstmLayer(6, "Name", "bilstm", "OutputMode", "last") % BiLSTM层 fullyConnectedLayer(1, "Name", "fc") % 全连接层 regressionLayer("Name", "regressionoutput")]; % 回归层 lgraph = addLayers(lgraph, tempLayers); % 将上述网络结构加入空白结构中 lgraph = connectLayers(lgraph, "seqfold/out", "conv_1"); % 折叠层输出 连接 卷积层输入; lgraph = connectLayers(lgraph, "seqfold/miniBatchSize", "sequnfold/miniBatchSize"); % 折叠层输出 连接 反折叠层输入 lgraph = connectLayers(lgraph, "conv_1", "relu_1"); % 卷积层输出 链接 激活层 lgraph = connectLayers(lgraph, "conv_1", "gapool"); % 卷积层输出 链接 全局平均池化 lgraph = connectLayers(lgraph, "relu_2", "multiplication/in2"); % 激活层输出 链接 相乘层 lgraph = connectLayers(lgraph, "sigmoid", "multiplication/in1"); % 全连接输出 链接 相乘层 lgraph = connectLayers(lgraph, "multiplication", "sequnfold/in"); % 点乘输出 %% 参数设置 options = trainingOptions('adam', ... % Adam 梯度下降算法 'MaxEpochs', 1000, ... % 最大迭代次数 'InitialLearnRate', 1e-2, ... % 初始学习率为0.01 'LearnRateSchedule', 'piecewise', ... % 学习率下降 'LearnRateDropFactor', 0.1, ... % 学习率下降因子 0.5 'LearnRateDropPeriod', 700, ... % 经过700次训练后 学习率为 0.01 * 0.1 'Shuffle', 'every-epoch', ... % 每次训练打乱数据集 'Plots', 'training-progress', ... % 画出曲线 'Verbose', false); %% 训练模型 net = trainNetwork(Lp_train, t_train, lgraph, options); %% 模型预测 t_sim1 = predict(net, Lp_train); t_sim2 = predict(net, Lp_test ); %% 数据反归一化 T_sim1 = mapminmax('reverse', t_sim1, ps_output); T_sim2 = mapminmax('reverse', t_sim2, ps_output); %% 均方根误差 error1 = sqrt(sum((T_sim1' - T_train).^2) ./ M); error2 = sqrt(sum((T_sim2' - T_test ).^2) ./ N); %% 显示网络结构 analyzeNetwork(net) %% 绘图 figure plot(1: M, T_train, 'r-*', 1: M, T_sim1, 'b-o', 'LineWidth', 1) legend('真实值', '预测值') xlabel('预测样本') ylabel('预测结果') string = {'训练集预测结果对比'; ['RMSE=' num2str(error1)]}; title(string) xlim([1, M]) grid figure plot(1: N, T_test, 'r-*', 1: N, T_sim2, 'b-o', 'LineWidth', 1) legend('真实值', '预测值') xlabel('预测样本') ylabel('预测结果') string = {'测试集预测结果对比'; ['RMSE=' num2str(error2)]}; title(string) xlim([1, N]) grid %% 相关指标计算 % R2 R1 = 1 - norm(T_train - T_sim1')^2 / norm(T_train - mean(T_train))^2; R2 = 1 - norm(T_test - T_sim2')^2 / norm(T_test - mean(T_test ))^2; disp(['训练集数据的R2为：', num2str(R1)]) disp(['测试集数据的R2为：', num2str(R2)]) % MAE mae1 = sum(abs(T_sim1' - T_train)) ./ M ; mae2 = sum(abs(T_sim2' - T_test )) ./ N ; disp(['训练集数据的MAE为：', num2str(mae1)]) disp(['测试集数据的MAE为：', num2str(mae2)]) % MBE mbe1 = sum(T_sim1' - T_train) ./ M ; mbe2 = sum(T_sim2' - T_test ) ./ N ; disp(['训练集数据的MBE为：', num2str(mbe1)]) disp(['测试集数据的MBE为：', num2str(mbe2)])