基于BP神经网络的多输出数据回归预测（Matlab完整程序和数据）

%% 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'); %% 划分训练集和测试集 temp = randperm(600); P_train = res(temp(1: 500), 1 : 28)'; T_train = res(temp(1: 500), 29: 31)'; M = size(P_train, 2); P_test = res(temp(501: end), 1 : 28)'; T_test = res(temp(501: end), 29: 31)'; 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); %% 创建网络 net = newff(p_train, t_train, 10); %% 设置训练参数 net.trainParam.epochs = 1000; % 迭代次数 net.trainParam.goal = 1e-6; % 误差阈值 net.trainParam.lr = 0.01; % 学习率 net.trainFcn = 'trainlm'; %% 训练网络 net = train(net, p_train, t_train); %% 仿真测试 t_sim1 = sim(net, p_train); t_sim2 = sim(net, p_test ); %% 数据反归一化 T_sim1 = mapminmax('reverse', t_sim1, ps_output); T_sim2 = mapminmax('reverse', t_sim2, ps_output); for i = 1: 3 %% 均方根误差 error1(i, :) = sqrt(sum((T_sim1(i, :) - T_train(i, :)).^2) ./ M); error2(i, :) = sqrt(sum((T_sim2(i, :) - T_test (i, :)).^2) ./ N); %% 绘图 figure subplot(2, 1, 1) plot(1: M, T_train(i, :), 'r-*', 1: M, T_sim1(i, :), '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(i, :))]}; title(string) xlim([1, M]) grid subplot(2, 1, 2) plot(1: N, T_test(i, :), 'r-*', 1: N, T_sim2(i, :), '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(i, :))]}; title(string) xlim([1, N]) grid %% 分割线 disp('**************************') disp(['下列是输出', num2str(i)]) disp('**************************') %% 相关指标计算 % 决定系数 R2 R1(i, :) = 1 - norm(T_train(i, :) - T_sim1(i, :))^2 / norm(T_train(i, :) - mean(T_train(i, :)))^2; R2(i, :) = 1 - norm(T_test (i, :) - T_sim2(i, :))^2 / norm(T_test (i, :) - mean(T_test (i, :)))^2; disp(['训练集数据的R2为：', num2str(R1(i, :))]) disp(['测试集数据的R2为：', num2str(R2(i, :))]) % 平均绝对误差 MAE mae1(i, :) = sum(abs(T_sim1(i, :) - T_train(i, :))) ./ M ; mae2(i, :) = sum(abs(T_sim2(i, :) - T_test (i, :))) ./ N ; disp(['训练集数据的MAE为：', num2str(mae1(i, :))]) disp(['测试集数据的MAE为：', num2str(mae2(i, :))]) % 平均相对误差 MBE mbe1(i, :) = sum(T_sim1(i, :) - T_train(i, :)) ./ M ; mbe2(i, :) = sum(T_sim2(i, :) - T_test (i, :)) ./ N ; disp(['训练集数据的MBE为：', num2str(mbe1(i, :))]) disp(['测试集数据的MBE为：', num2str(mbe2(i, :))]) end