【故障诊断】基于KNN、SVM、RF、DT、ET多种算法实现制冷系统故障诊断附Matlab代码...

✅作者简介：热爱科研的Matlab仿真开发者，修心和技术同步精进，matlab项目合作可私信。

🍎个人主页：Matlab科研工作室

🍊个人信条：格物致知。

⛄ 内容介绍

制冷系统故障可由多种模型进行模拟诊断.为了提高其诊断性能,将包括K近邻模型(KNN),支持向量机(SVM),决策树模型(DT),随机森林模型(RF)及逻辑斯谛回归模型(LR)在内的5种成员诊断器,通过绝对多数投票方法集成为一个集成模型,并采用美国采暖,制冷与空调工程师学会(ASHRAE)故障数据对1台90冷吨的离心式冷水机组进行建模及验证,数据包含制冷系统的7类典型故障及一类正常运行.结果表明:集成模型在所选数据集上总体诊断正确率达到99.58%,较各成员诊断器(94.55%~99.05%)均有显著提升,对正常运行,局部故障及全局故障的诊断性能亦有改善.此外,对比分析了不同集成模型及成员诊断器的诊断性能,从中找到诊断正确率与时间成本最佳的集成模型(99.41%,1.34 s).可见,集成模型较单一模型性能更佳,在制冷系统故障诊断中具有更好的应用前景.

⛄ 部分代码

% Gaussian Mixture Model (10/12/2020)

function GMM = mGaussianMixtureModel(feat,label,opts)
% Default
kfold = 10;
tf    = 2;

if isfield(opts,'kfold'), kfold = opts.kfold; end
if isfield(opts,'ho'), ho = opts.ho; end
if isfield(opts,'tf'), tf = opts.tf; end

% Number of class
num_class = numel(unique(label)); 

% [Hold-out]
if tf == 1
  fold = cvpartition(label,'HoldOut',ho);
  % Call train & test data
  xtrain = feat(fold.training,:); ytrain = label(fold.training);
  xtest  = feat(fold.test,:);     ytest2 = label(fold.test);
  % Train model
  My_Model = fitgmdist(xtrain,num_class,...
    'Options',statset('MaxIter',1000),...
    'Regularize',1e-5,...
    'Start',ytrain);
  % Test using cluster
  pred2 = cluster(My_Model,xtest); 
  % Accuracy
  Afold = sum(pred2 == ytest2) / length(ytest2);  
  
% [Cross-validation] 
elseif tf == 2
  fold   = cvpartition(label,'KFold',kfold);
  Afold  = zeros(kfold,1); 
  pred2  = [];
  ytest2 = []; 
  for i = 1:kfold
    % Call train & test data
    trainIdx = fold.training(i); testIdx = fold.test(i);
    xtrain   = feat(trainIdx,:); ytrain  = label(trainIdx);
    xtest    = feat(testIdx,:);  ytest   = label(testIdx); 
    % Train model
    My_Model = fitgmdist(xtrain,num_class,...
      'Options',statset('MaxIter',1000),...
      'Regularize',1e-5,...
      'Start',ytrain);
    % Test using cluster
    pred = cluster(My_Model,xtest); 
    % Accuracy
    Afold(i) = sum(pred == ytest) / length(ytest);
    % Store temporary
    pred2  = [pred2(1:end); pred];
    ytest2 = [ytest2(1:end); ytest]; 
  end
  
% [Leave-one out]
elseif tf == 3
  fold     = cvpartition(label,'LeaveOut');
  % Size of data
  num_data = length(label); 
  Afold    = zeros(num_data,1); 
  pred2    = [];
  ytest2   = []; 
  for i = 1:num_data
    % Call train & test data
    trainIdx = fold.training(i); testIdx = fold.test(i);
    xtrain   = feat(trainIdx,:); ytrain  = label(trainIdx);
    xtest    = feat(testIdx,:);  ytest   = label(testIdx); 
    % Train model
    My_Model = fitgmdist(xtrain,num_class,...
      'Options',statset('MaxIter',1000),...
      'Regularize',1e-5,...
      'Start',ytrain);
    % Test using cluster
    pred = cluster(My_Model,xtest); 
    % Accuracy
    Afold(i) = sum(pred == ytest) / length(ytest);
    % Store temporary
    pred2  = [pred2(1:end); pred];
    ytest2 = [ytest2(1:end); ytest]; 
  end
end
% Confusion matrix
confmat = confusionmat(ytest2,pred2); 
% Overall accuracy
acc = mean(Afold); 
% Store result
GMM.acc  = acc;
GMM.con  = confmat;

if tf == 1
  fprintf('\n Accuracy (GMM-HO): %g %%',100 * acc);
elseif tf == 2
  fprintf('\n Accuracy (GMM-CV): %g %%',100 * acc);
elseif tf == 3
  fprintf('\n Accuracy (GMM-LOO): %g %%',100 * acc);
end
end

⛄ 运行结果

⛄ 参考文献

Katırcıoğlu F, Cingiz Z. Fault diagnosis for overcharge and undercharge conditions in refrigeration systems using infrared thermal images. Proceedings of the Institution of Mechanical Engineers, Part E: Journal of Process Mechanical Engineering. 2023;0(0). doi:10.1177/09544089221148065

⛳️ 代码获取关注我

❤️部分理论引用网络文献，若有侵权联系博主删除

❤️ 关注我领取海量matlab电子书和数学建模资料