机器学习之MATLAB代码–基于VMD与SSA优化lssvm的功率预测(多变量)(七)

  • 代码
    • 数据
      • 结果

代码

先对外层代码的揭露,包括:顺序而下

1、

function s = Bounds( s, Lb, Ub)% Apply the lower bound vectortemp = s;I = temp < Lb;temp(I) = Lb(I);% Apply the upper bound vectorJ = temp > Ub;temp(J) = Ub(J);% Update this new moves = temp;

2、

function [in,out]=data_process(data,num)% 采用1-num的各种值为输入 第num+1的功率作为输出n=length(data)-num;for i=1:nx(i,:,:)=data(i:i+num,:);endin=x(1:end-1,:);out=x(2:end,end);%功率是最后一列

3、

function y=fitness(x,train_x,train_y,test_x,test_y,typeID,kernelnamescell)gam=x(1);sig2=x(2);model=initlssvm(train_x,train_y,typeID,gam,sig2,kernelnamescell);model=trainlssvm(model);y_pre_test=simlssvm(model,test_x);y=mse(test_y-y_pre_test);

4、

clc;clear;close all%%lssvm=load('lssvm.mat');ssa_lssvm=load('ssa_lssvm.mat');vmd_lssvm=load('vmd_lssvm.mat');vmd_ssa_lssvm=load('vmd_ssa_lssvm.mat');disp('结果分析-lssvm')result(lssvm.true_value,lssvm.predict_value)fprintf('\n')disp('结果分析-ssa-lssvm')result(ssa_lssvm.true_value,ssa_lssvm.predict_value)fprintf('\n')disp('结果分析-vmd-lssvm')result(vmd_lssvm.true_value,vmd_lssvm.predict_value)fprintf('\n')disp('结果分析-vmd-ssa-lssvm')result(vmd_ssa_lssvm.true_value,vmd_ssa_lssvm.predict_value)figureplot(lssvm.true_value)hold on;grid onplot(lssvm.predict_value)plot(ssa_lssvm.predict_value)plot(vmd_lssvm.predict_value)plot(vmd_ssa_lssvm.predict_value)legend('真实值','lssvm预测','ssa-lssvm预测','vmd-lssvm预测','vmd-ssa-lssvm预测')title('各算法预测效果对比')ylabel('功率值')xlabel('测试集样本')

5、

%% 其他数据与功率数据一起作为输入,参看data_process.mclc;clear;close alladdpath(genpath('LSSVMlabv1_8'));%%data=xlsread('预测数据.xls','B2:K1000');[x,y]=data_process(data,24);%24个时刻 预测下一个时刻%归一化[xs,mappingx]=mapminmax(x',0,1);x=xs';[ys,mappingy]=mapminmax(y',0,1);y=ys';%划分数据n=size(x,1);m=round(n*0.7);%70%训练,对最后30%进行预测train_x=x(1:m,:);test_x=x(m+1:end,:);train_y=y(1:m,:);test_y=y(m+1:end,:);%% lssvm 参数gam=10;sig2=1000;typeID='function estimation';kernelnamescell='RBF_kernel';model=initlssvm(train_x,train_y,typeID,gam,sig2,kernelnamescell);model=trainlssvm(model);y_pre_test=simlssvm(model,test_x);% 反归一化predict_value=mapminmax('reverse',y_pre_test',mappingy);true_value=mapminmax('reverse',test_y',mappingy);save lssvm predict_value true_valuedisp('结果分析')rmse=sqrt(mean((true_value-predict_value).^2));disp(['根均方差(RMSE):',num2str(rmse)])mae=mean(abs(true_value-predict_value));disp(['平均绝对误差(MAE):',num2str(mae)])mape=mean(abs(true_value-predict_value)/true_value);disp(['平均相对百分误差(MAPE):',num2str(mape*100),'%'])fprintf('\n')figureplot(true_value,'-*','linewidth',3)hold onplot(predict_value,'-s','linewidth',3)legend('实际值','预测值')grid ontitle('LSSVM')

6、

clc;clear;close all;format compactaddpath(genpath('LSSVMlabv1_8'));%%data=xlsread('预测数据.xls','B2:K1000');[x,y]=data_process(data,24);%24个时刻 预测下一个时刻%归一化[xs,mappingx]=mapminmax(x',0,1);x=xs';[ys,mappingy]=mapminmax(y',0,1);y=ys';%划分数据n=size(x,1);m=round(n*0.7);%70%训练,对最后30%进行预测train_x=x(1:m,:);test_x=x(m+1:end,:);train_y=y(1:m,:);test_y=y(m+1:end,:);%% ssa优化lssvm 参数typeID='function estimation';kernelnamescell='RBF_kernel';[x,trace]=ssa_lssvm(typeID,kernelnamescell,train_x,train_y,test_x,test_y);figure;plot(trace);title('适应度曲线/mse')%% 利用寻优得到的参数重新训练lssvmdisp('寻优得到的参数分别是:')gam=x(1)sig2=x(2)model=initlssvm(train_x,train_y,typeID,gam,sig2,kernelnamescell);model=trainlssvm(model);y_pre_test=simlssvm(model,test_x);% 反归一化predict_value=mapminmax('reverse',y_pre_test',mappingy);true_value=mapminmax('reverse',test_y',mappingy);save ssa_lssvm predict_value true_valuedisp('结果分析')rmse=sqrt(mean((true_value-predict_value).^2));disp(['根均方差(RMSE):',num2str(rmse)])mae=mean(abs(true_value-predict_value));disp(['平均绝对误差(MAE):',num2str(mae)])mape=mean(abs(true_value-predict_value)/true_value);disp(['平均相对百分误差(MAPE):',num2str(mape*100),'%']) fprintf('\n')figureplot(true_value,'-*','linewidth',3)hold onplot(predict_value,'-s','linewidth',3)legend('实际值','预测值')grid ontitle('SSA-LSSVM')

7、

clc;clear;format compact;close all;ticrng('default')%% 数据预处理data=xlsread('预测数据.xls','B2:K1000');figure;plot(data(:,end),'-*','linewidth',3);title('原始功率曲线');grid on;ylabel('功率值')alpha = 2000;% moderate bandwidth constrainttau = 0;% noise-tolerance (no strict fidelity enforcement)K = 5;% 3 modesDC = 0; % no DC part imposedinit = 1; % initialize omegas uniformlytol = 1e-7;[imf, u_hat, omega] = VMD(data(:,end), alpha, tau, K, DC, init, tol);figurefor i =1:size(imf,1)subplot(size(imf,1),1,i)plot(imf(i,:))ylabel(['imf',num2str(i)])endylabel('残余')suptitle('VMD')c=size(imf,1);pre_result=[];true_result=[];%% 对每个分量建模for i=1:cdisp(['对第',num2str(i),'个分量建模'])[x,y]=data_process([data(:,1:end-1) imf(i,:)'],24);%每个序列和原始的几列数据合并 然后划分%归一化[xs,mappingx]=mapminmax(x',0,1);x=xs';[ys,mappingy]=mapminmax(y',0,1);y=ys';%划分数据n=size(x,1);m=round(n*0.7);%70%训练,对最后30%进行预测train_x=x(1:m,:);test_x=x(m+1:end,:);train_y=y(1:m,:);test_y=y(m+1:end,:);%%gam=100;sig2=1000;typeID='function estimation';kernelnamescell='RBF_kernel';model=initlssvm(train_x,train_y,typeID,gam,sig2,kernelnamescell);model=trainlssvm(model);pred=simlssvm(model,test_x);pre_value=mapminmax('reverse',pred',mappingy);true_value=mapminmax('reverse',test_y',mappingy);pre_result=[pre_result;pre_value];true_result=[true_result;true_value];end%% 各分量预测的结果相加true_value=sum(true_result);predict_value=sum(pre_result);save vmd_lssvm predict_value true_value%%load vmd_lssvmdisp('结果分析')rmse=sqrt(mean((true_value-predict_value).^2));disp(['根均方差(RMSE):',num2str(rmse)])mae=mean(abs(true_value-predict_value));disp(['平均绝对误差(MAE):',num2str(mae)])mape=mean(abs(true_value-predict_value)/true_value);disp(['平均相对百分误差(MAPE):',num2str(mape*100),'%'])fprintf('\n')figureplot(true_value,'-*','linewidth',3)hold onplot(predict_value,'-s','linewidth',3)legend('实际值','预测值')grid ontitle('VMD-LSSVM')

8、

clc;clear;format compact;close all;ticrng('default')%% 数据预处理data=xlsread('预测数据.xls','B2:K1000');figure;plot(data(:,end),'-*','linewidth',3);title('原始功率曲线');grid on;ylabel('功率值')alpha = 2000;% moderate bandwidth constrainttau = 0;% noise-tolerance (no strict fidelity enforcement)K = 5;% 3 modesDC = 0; % no DC part imposedinit = 1; % initialize omegas uniformlytol = 1e-7;[imf, u_hat, omega] = VMD(data(:,end), alpha, tau, K, DC, init, tol);figurefor i =1:size(imf,1)subplot(size(imf,1),1,i)plot(imf(i,:))ylabel(['imf',num2str(i)])endylabel('残余')suptitle('VMD')c=size(imf,1);pre_result=[];true_result=[];%% 对每个分量建模for i=1:cdisp(['对第',num2str(i),'个分量建模'])[x,y]=data_process([data(:,1:end-1) imf(i,:)'],24);%归一化[xs,mappingx]=mapminmax(x',0,1);x=xs';[ys,mappingy]=mapminmax(y',0,1);y=ys';%划分数据n=size(x,1);m=round(n*0.7);%70%训练,对最后30%进行预测train_x=x(1:m,:);test_x=x(m+1:end,:);train_y=y(1:m,:);test_y=y(m+1:end,:);%%typeID='function estimation';kernelnamescell='RBF_kernel';[x,trace]=ssa_lssvm(typeID,kernelnamescell,train_x,train_y,test_x,test_y);gam=x(1);sig2=x(2);model=initlssvm(train_x,train_y,typeID,gam,sig2,kernelnamescell);model=trainlssvm(model);pred=simlssvm(model,test_x);pre_value=mapminmax('reverse',pred',mappingy);true_value=mapminmax('reverse',test_y',mappingy);pre_result=[pre_result;pre_value];true_result=[true_result;true_value];end%% 各分量预测的结果相加true_value=sum(true_result);predict_value=sum(pre_result);save vmd_ssa_lssvm predict_value true_value%%load vmd_ssa_lssvmdisp('结果分析')rmse=sqrt(mean((true_value-predict_value).^2));disp(['根均方差(RMSE):',num2str(rmse)])mae=mean(abs(true_value-predict_value));disp(['平均绝对误差(MAE):',num2str(mae)])mape=mean(abs(true_value-predict_value)/true_value);disp(['平均相对百分误差(MAPE):',num2str(mape*100),'%'])fprintf('\n')figureplot(true_value,'-*','linewidth',3)hold onplot(predict_value,'-s','linewidth',3)legend('实际值','预测值')grid ontitle('VMD-SSA-LSSVM')

9、

clear;close all;clc;format compactaddpath('vmd-verify')data=xlsread('预测数据.xls','B2:K1000');f=data(:,end);% some sample parameters for VMDalpha = 2000;% moderate bandwidth constrainttau = 0;% noise-tolerance (no strict fidelity enforcement)K = 5;% 3 modesDC = 0; % no DC part imposedinit = 1; % initialize omegas uniformlytol = 1e-7;%--------------- Run actual VMD code[u, u_hat, omega] = VMD(f, alpha, tau, K, DC, init, tol);figureplot(f)title('原始')figurefor i=1:Ksubplot(K,1,i)plot(u(i,:))ylabel(['IMF_',num2str(i)])endylabel('res')

10、

function result(true_value,predict_value)rmse=sqrt(mean((true_value-predict_value).^2));disp(['根均方差(RMSE):',num2str(rmse)])mae=mean(abs(true_value-predict_value));disp(['平均绝对误差(MAE):',num2str(mae)])mape=mean(abs(true_value-predict_value)/true_value);disp(['平均相对百分误差(MAPE):',num2str(mape*100),'%'])

11、

function [bestX,Convergence_curve]=ssa_lssvm(typeID,Kernel_type,inputn_train,label_train,inputn_test,label_test)%% 麻雀优化pop=10; % 麻雀数M=10; % Maximum numbef of iterationsc=1;d=10000;dim=2;P_percent = 0.2;% The population size of producers accounts for "P_percent" percent of the total population size%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%pNum = round( pop *P_percent );% The population size of the producerslb= c.*ones( 1,dim );% Lower limit/bounds/ a vectorub= d.*ones( 1,dim );% Upper limit/bounds/ a vector%Initializationfor i = 1 : popx( i, : ) = lb + (ub - lb) .* rand( 1, dim );fit( i )=fitness(x(i,:),inputn_train,label_train,inputn_test,label_test,typeID,Kernel_type); endpFit = fit;pX = x;% The individual's best position corresponding to the pFit[ fMin, bestI ] = min( fit );% fMin denotes the global optimum fitness valuebestX = x( bestI, : ); % bestX denotes the global optimum position corresponding to fMinfor t = 1 : M[ ans, sortIndex ] = sort( pFit );% Sort.[fmax,B]=max( pFit );worse= x(B,:);r2=rand(1);%%%%%%%%%%%%%5%%%%%%这一部位为发现者(探索者)的位置更新%%%%%%%%%%%%%%%%%%%%%%%%%if(r2<0.8)%预警值较小,说明没有捕食者出现for i = 1 : pNum%r2小于0.8的发现者的改变(1-20% Equation (3)r1=rand(1);x( sortIndex( i ), : ) = pX( sortIndex( i ), : )*exp(-(i)/(r1*M));%对自变量做一个随机变换x( sortIndex( i ), : ) = Bounds( x( sortIndex( i ), : ), lb, ub );%对超过边界的变量进行去除fit( sortIndex( i ) )=fitness(x(sortIndex( i ),:),inputn_train,label_train,inputn_test,label_test,typeID,Kernel_type); endelse %预警值较大,说明有捕食者出现威胁到了种群的安全,需要去其它地方觅食for i = 1 : pNum %r2大于0.8的发现者的改变x( sortIndex( i ), : ) = pX( sortIndex( i ), : )+randn(1)*ones(1,dim);x( sortIndex( i ), : ) = Bounds( x( sortIndex( i ), : ), lb, ub );fit( sortIndex( i ) )=fitness(x(sortIndex( i ),:),inputn_train,label_train,inputn_test,label_test,typeID,Kernel_type); endend[ fMMin, bestII ] = min( fit );bestXX = x( bestII, : );%%%%%%%%%%%%%5%%%%%%这一部位为加入者(追随者)的位置更新%%%%%%%%%%%%%%%%%%%%%%%%%for i = ( pNum + 1 ) : pop %剩下20-100的个体的变换% Equation (4) A=floor(rand(1,dim)*2)*2-1;if( i>(pop/2))%这个代表这部分麻雀处于十分饥饿的状态(因为它们的能量很低,也是是适应度值很差),需要到其它地方觅食x( sortIndex(i ), : )=randn(1)*exp((worse-pX( sortIndex( i ), : ))/(i)^2);else%这一部分追随者是围绕最好的发现者周围进行觅食,其间也有可能发生食物的争夺,使其自己变成生产者x( sortIndex( i ), : )=bestXX+(abs(( pX( sortIndex( i ), : )-bestXX)))*(A'*(A*A')^(-1))*ones(1,dim);endx( sortIndex( i ), : ) = Bounds( x( sortIndex( i ), : ), lb, ub );%判断边界是否超出fit( sortIndex( i ) )=fitness(x(sortIndex( i ),:),inputn_train,label_train,inputn_test,label_test,typeID,Kernel_type); end%%%%%%%%%%%%%5%%%%%%这一部位为意识到危险(注意这里只是意识到了危险,不代表出现了真正的捕食者)的麻雀的位置更新%%%%%%%%%%%%%%%%%%%%%%%%%c=randperm(numel(sortIndex));%%%%%%%%%这个的作用是在种群中随机产生其位置(也就是这部分的麻雀位置一开始是随机的,意识到危险了要进行位置移动,%处于种群外围的麻雀向安全区域靠拢,处在种群中心的麻雀则随机行走以靠近别的麻雀)b=sortIndex(c(1:10));for j =1: length(b)% Equation (5)if( pFit( sortIndex( b(j) ) )>(fMin) ) %处于种群外围的麻雀的位置改变x( sortIndex( b(j) ), : )=bestX+(randn(1,dim)).*(abs(( pX( sortIndex( b(j) ), : ) -bestX)));else %处于种群中心的麻雀的位置改变x( sortIndex( b(j) ), : ) =pX( sortIndex( b(j) ), : )+(2*rand(1)-1)*(abs(pX( sortIndex( b(j) ), : )-worse))/ ( pFit( sortIndex( b(j) ) )-fmax+1e-50);endx( sortIndex(b(j) ), : ) = Bounds( x( sortIndex(b(j) ), : ), lb, ub );fit( sortIndex( b(j) ) )=fitness(x(sortIndex(b(j) ),:),inputn_train,label_train,inputn_test,label_test,typeID,Kernel_type); endfor i = 1 : popif ( fit( i ) < pFit( i ) )pFit( i ) = fit( i );pX( i, : ) = x( i, : );endif( pFit( i ) < fMin )fMin= pFit( i );bestX = pX( i, : );endendConvergence_curve(t,:)=[fMin mean(pFit)];end

接下来是内嵌代码,就是下面两个文件夹的代码,实在是太多,想要的留言吧!

数据

数据是由时间、风速、风向等等因素组成的文件。

结果






结果图太多,就先给出这么多,如有需要代码和数据的同学请在评论区发邮箱,一般一天之内会回复,请点赞+关注谢谢!!