MATLAB实现PSO-DBN粒子群算法优化深度置信网络多输入单输出回归预测（完整源码和数据）

%% 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'); %% 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); %% 转置以适应模型 p_train = p_train'; p_test = p_test'; t_train = t_train'; t_test = t_test'; %% 预训练的参数设置 opts.numepochs = 500; % 训练次数 opts.batchsize = 12; % 每次训练样本个数 需满足：（M / batchsize = 整数） opts.momentum = 0; % 学习率的动量 opts.alpha = 0.05; % 学习率 %% 粒子群的参数初始化 c1 = 54.95; % 学习因子 c2 = 4.494; % 学习因子 maxgen = 30; % 种群更新次数 sizepop = 5; % 种群规模 %% 设置边界 % ------ 前 numsum 个数为隐藏层节点边界和速度，最后一个为反向学习率，倒数第二个为迭代次数 ------- Vmax = [ 20, 20, 400, 1.0]; % 最大速度 Vmin = [ -5, -5, -50, -0.5]; % 最小速度 popmax = [100, 100, 5000, 4.0]; % 最大边界 popmin = [ 10, 10, 200, 0.5]; % 最小边界 %% 节点总数 numsum = length(popmax) - 2; % 隐藏层层数 for i = 1 : sizepop pop(i, :) = rand(1, length(popmax)) .* (popmax - popmin) ./ 1.2 + popmax ./ 100; V(i, :) = rand(1, length(Vmax)) .* Vmax + 1; fitness(i) = fun(pop(i, :), numsum, p_train, t_train, opts); end %% 个体极值和群体极值 [fitnesszbest, bestindex] = min(fitness); zbest = pop(bestindex, :); % 全局最佳 gbest = pop; % 个体最佳 fitnessgbest = fitness; % 个体最佳适应度值 BestFit = fitnesszbest; % 全局最佳适应度值 %% 迭代寻优 for i = 1 : maxgen for j = 1 : sizepop % 速度更新 V(j, :) = V(j, :) + c1 * rand * (gbest(j, :) - pop(j, :)) + c2 * rand * (zbest - pop(j, :)); if (sum(V(j, :) > Vmax) > 0 || sum(V(j, :) < Vmin) > 0) [~, vmax_index] = find(V(j, :) > Vmax); [~, vmin_index] = find(V(j, :) < Vmin); V(j, vmax_index) = Vmax(vmax_index); V(j, vmin_index) = Vmin(vmin_index); end % 种群更新 pop(j, :) = pop(j, :) + 0.2 * V(j, :); if (sum(pop(j, :) > popmax) > 0 || sum(pop(j, :) < popmin) > 0) [~, pmax_index] = find(pop(j, :) > popmax); [~, pmin_index] = find(pop(j, :) < popmin); pop(j, pmax_index) = popmax(pmax_index); pop(j, pmin_index) = popmin(pmin_index); end % 自适应变异 if rand > 0.85 pop(j, :) = rand(1, length(popmax)) .* (popmax - popmin) / 1.2 + popmax ./ 100; end % 适应度值 fitness(j) = fun(pop(j, :), numsum, p_train, t_train, opts); end for j = 1 : sizepop % 个体最优更新 if fitness(j) < fitnessgbest(j) gbest(j, :) = pop(j, :); fitnessgbest(j) = fitness(j); end % 群体最优更新 if fitness(j) < fitnesszbest zbest = pop(j, :); fitnesszbest = fitness(j); end end BestFit = [BestFit, fitnesszbest]; end %% 提取最优参数 for i = 1 : numsum dbn.sizes(i + 1) = round(zbest(i)); end dbn.sizes(1) = []; % lr = zbest(end); % 学习率 epochs = round(zbest(numsum + 1)); % 迭代次数 %% 训练模型 dbn = dbnsetup(dbn, p_train, opts); % 建立模型 dbn = dbntrain(dbn, p_train, opts); % 训练模型 %% 训练权重移植，添加输出层 nn = dbnunfoldtonn(dbn, outdim); %% 反向调整网络 % ----- 参数修改时，请查看fun函数中的参数设置，并作出对应修改 --------- opts.numepochs = epochs; % 反向微调次数 opts.batchsize = 12; % 每次反向微调样本数 需满足：（M / batchsize = 整数） nn.activation_function = 'sigm'; % 激活函数 nn.learningRate = lr; % 学习率 nn.momentum = 0.5; % 动量参数 nn.scaling_learningRate = 1; % 学习率的比例因子 [nn, loss] = nntrain(nn, p_train, t_train, opts); % 训练 %% 预测 t_sim1 = nnpredict(nn, p_train); t_sim2 = nnpredict(nn, p_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); %% 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 %% 绘制损失函数曲线 figure plot(1 : length(loss), loss, 'b-', 'LineWidth', 1) xlim([1, length(loss)]) xlabel('迭代次数') ylabel('误差损失') legend('损失函数') title('损失函数') grid %% 误差曲线迭代图 figure; plot(1 : length(BestFit), BestFit, 'LineWidth', 1.5); xlabel('粒子群迭代次数'); ylabel('适应度值'); xlim([1, length(BestFit)]) string = {'模型迭代误差变化'}; title(string) grid on %% 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)])