👨🎓个人主页:研学社的博客
💥💥💞💞欢迎来到本博客❤️❤️💥💥
🏆博主优势:🌞🌞🌞博客内容尽量做到思维缜密,逻辑清晰,为了方便读者。
⛳️座右铭:行百里者,半于九十。
📋📋📋本文目录如下:🎁🎁🎁
目录
针对目前非侵入式负荷监测仅能识别单个家用电器、多种家用电器同时运行识别率低的问题,本文提出了一种提取电流波形特征以识别家用电器的方法。使用短时傅里叶变换(STFT)和核PCA技术来提取这些特征。定义特征后,分类器k-最近邻(kNN),支持向量机(SVM),线性判别分析(LDA),随机森林(RF)和极限学习机(ELM)用于设备(或设备组合)识别。
部分代码:
%% Classification
%% KNN
% Train
start_time_train=cputime;
KNN = fitcknn(Xtrainp,Ytrain,'NumNeighbors',5);
end_time_train=cputime;
TrainTime_knn=end_time_train-start_time_train
% Test
start_time_test=cputime;
label_knn = predict(KNN,Xtestp);
end_time_test=cputime;
TestTime_knn=end_time_test-start_time_test
AccRate_Knn = sum(Ytest' == label_knn)/length(Ytest)
%% SVM
% Train
start_time_train=cputime;
tSVM = templateSVM('Standardize',1,'KernelFunction','linear');
SVM = fitcecoc(Xtrainp,Ytrain,'Learners',tSVM);
end_time_train=cputime;
TrainTime_svm=end_time_train-start_time_train
% Test
start_time_test=cputime;
label_svm = predict(SVM,Xtestp);
end_time_test=cputime;
TestTime_svm=end_time_test-start_time_test
AccRat_svm = sum(Ytest' == label_svm)/length(Ytest)
%% LDA
% Train
start_time_train=cputime;
LDA = fitcdiscr(Xtrainp,Ytrain);
end_time_train=cputime;
TrainTime_lda=end_time_train-start_time_train
% Test
start_time_test=cputime;
label_lda = predict(LDA,Xtestp);
end_time_test=cputime;
TestTime_lda=end_time_test-start_time_test
AccRate_lda = sum(Ytest' == label_lda)/length(Ytest)
%% Decision tree
% train
start_time_train=cputime;
TREE = TreeBagger(150,Xtrainp,Ytrain,'OOBPredictorImportance','on');
end_time_train=cputime;
TrainTime_tree=end_time_train-start_time_train
% Test
start_time_test=cputime;
label_tree = str2double(predict(TREE,Xtestp));
end_time_test=cputime;
TestTime_tree=end_time_test-start_time_test
AccRate_tree = sum(Ytest' == label_tree)/length(Ytest)
%% ELM
[TrainingTime, TestingTime, ~, AccRate_elm, label_elm] = elm_aece(Xtrainp',...
Ytrain, Xtestp', Ytest, 1, 1500, 'sig');
AccRate_elm
%% Statistics
C_svm = confusionmat(Ytest',label_svm);
C_lda = confusionmat(Ytest',label_lda);
C_knn = confusionmat(Ytest',label_knn);
C_tree = confusionmat(Ytest',label_tree);
C_elm = confusionmat(Ytest',label_elm);
[result_svm,reference_svm]= getStatistical(C_svm);
[result_knn,reference_knn]= getStatistical(C_knn);
[result_lda,reference_lda]= getStatistical(C_lda);
[result_tree,reference_tree]= getStatistical(C_tree);
[result_elm,reference_elm]= getStatistical(C_elm);
部分理论来源于网络,如有侵权请联系删除。
[1]李争,王泽,冯威,安国庆,王强,陈贺.基于CNN与K-means聚类的非侵入式电器负荷识别方法[J].河北科技大学学报,2022,43(04):365-373.
[2]Daniel Cavalieri (2022). KPCA and STFT Non-intrusive Load Monitoring