mdl =“pdmTransmissionCasing”；open_system (mdl)

open_system ([mdl' /套管'])

open_system ([mdl/带漂移的振动传感器])

open_system ([mdl“/轴”])

(mdl open_system (,/轴输出轴的])

toothFaultArray = -2:2/10:0;%牙齿故障增益值sensorDriftArray = -1:0.5:1;%传感器漂移偏移值shaftWearArray = [0 -1];%可用于驱动轴条件的变体用所有值的组合创建一个n维数组[toothFaultValues, sensorDriftValues shaftWearValues] =.．.ndgrid (toothFaultArray sensorDriftArray shaftWearArray);为ct = numel(toothFaultValues):-1:1创建一个Simulin金宝appk。SimulationInput为each combination of valuessiminput = 金宝appSimulink.SimulationInput(mdl);修改模型参数siminput = setVariable(siminput，“ToothFaultGain”toothFaultValues (ct));siminput = setVariable(siminput，“SDrift”sensorDriftValues (ct));siminput = setVariable(siminput，“ShaftWear”shaftWearValues (ct));在数组中收集模拟输入gridSimulationInput(ct) = simminput;结束

rng (“默认”）;重置随机种子的重现性toothFaultArray = [0 -rand(1,6)];%牙齿故障增益值sensorDriftArray = [0 randn(1,6)/8];%传感器漂移偏移值shaftWearArray = [0 -1];%可用于驱动轴条件的变体用所有值的组合创建一个n维数组[toothFaultValues, sensorDriftValues shaftWearValues] =.．.ndgrid (toothFaultArray sensorDriftArray shaftWearArray);为ct =元素个数(toothFaultValues): 1:1创建一个Simulin金宝appk。SimulationInput为each combination of valuessiminput = 金宝appSimulink.SimulationInput(mdl);修改模型参数siminput = setVariable(siminput，“ToothFaultGain”toothFaultValues (ct));siminput = setVariable(siminput，“SDrift”sensorDriftValues (ct));siminput = setVariable(siminput，“ShaftWear”shaftWearValues (ct));在数组中收集模拟输入randomSimulationInput(ct) = siminput;结束

运行模拟并创建集成以管理模拟结果。如果~ (fullfile (pwd,存在“数据”)，“dir”) mkdir (fullfile (pwd,“数据”）)创建目录来存储结果结束runAll = true;如果runAll [ok,e] = generateSimulationEnsemble([gridSimulationInput, randomSimulationInput]，.．.fullfile (pwd,“数据”)，“UseParallel”,真正的);其他的[ok,e] = generateSimulationEnsemble(gridSimulationInput(1:10)， fullfile(pwd，“数据”));% #好< * UNRCH >结束

ens = simulationEnsembleDatastore(fullfile(pwd，“数据”));

实体

ens.SelectedVariables

ens.SelectedVariables = [“振动”“转速器”“SimulationInput”];数据=读取(ens)

振动=数据。振动{1};情节(vibration.Time vibration.Data)标题(“振动”) ylabel (“加速”）

Idx =振动。时间>=秒(10);振动=振动(idx，:);振动。时间=震动。Time -振动。Time(1);

tacho = data.Tacho{1};Idx = tacho。时间>=秒(10);Tacho = Tacho (idx，:);情节(tacho.Time tacho.Data)标题(转速计脉冲的)传说(驱动轴的，“负载轴”）负载轴比传动轴转动慢

idx = diff(tacho.Data(:，2)) > 0.5;tachoPulses = tacho.Time(find(idx)+1)-tacho.Time(1)

vars = data.SimulationInput{1}.Variables;Idx = strcmp({vars。名称},“SDrift”）;如果any(idx) sF = abs(vars(idx).Value) > 0.01;漂移值小不是故障其他的sF = false;结束Idx = strcmp({vars。名称},“ShaftWear”）;如果any(idx) sV = vars(idx)。值< 0;其他的sV = false;结束如果Any (idx) idx = strcmp({vars. txt)名称},“ToothFaultGain”）;sT = abs(vars(idx).Value) < 0.1;%小齿故障值不是故障其他的sT = false结束faultCode = sF + 2*sV + 4*sT;用于捕获不同故障条件的故障代码

sdata = table({振动}，{tachoPulses}，sF,sV,sT,faultCode，.．.“VariableNames”, {“振动”，“TachoPulses”，“SensorDrift”，“ShaftWear”，“ToothFault”，“FaultCode”})

ens.DataVariables = [ens.DataVariables;“TachoPulses”];

ens.ConditionVariables = [“SensorDrift”，“ShaftWear”，“ToothFault”，“FaultCode”];

reset(ens) runLocal = false;如果runLocal处理集成中的每个成员而Hasdata (ens) data = read(ens);addData = prepareData(data);writeToLastMemberRead(实体,addData)结束其他的将集成拆分为多个分区，并并行处理每个分区N = numpartitions(ens,gcp);parforCt = 1:n subens =分区(ens,n, Ct);而Hasdata (subens) data = read(subens);addData = prepareData(data);writeToLastMemberRead (suben, addData)结束结束结束

reset(ens) ens. selectedvariables =“振动”；图，ct = 1;而Hasdata (ens) data = read(ens);如果mod(ct,10) == 0振动=数据{1};情节(vibration.Time vibration.Data)在结束Ct = Ct + 1;结束持有从标题(振动信号的) ylabel (“加速”）

ens.SelectedVariables = [“振动”，“TachoPulses”];

Reset (ens) data = read(ens)

振动=数据。振动{1};将振动信号插值到适合fft分析的周期时间基上Np = 2^floor(log(高度(振动))/log(2));dt =振动。时间(结束)/(np-1);tv = 0:dt:振动。时间(结束);Y = retime(振动，电视，“线性”）;计算振动信号的时间同步平均值tp = seconds(data.TachoPulses{1});vibrationTSA = tsa(y,tp);图(vibrationTSA.ttTime,vibrationTSA.tsa)“振动时间同步平均”) ylabel (“加速”）

计算时间同步平均的频谱np =数字(振动tsa);f = fft(vibrationTSA.tsa.*hamming(np))/np;frTSA = f(1:floor(np/2)+1);% TSA频率响应wTSA = (0:np/2)/np*(2*pi/seconds(dt));TSA频谱频率mTSA = abs(frTSA);% TSA谱量级图半对数(wTSA,20*log10(mTSA))振动光谱的)包含(“rad / s”）

reset(ens) ens. datavariables = [ens. datavariables;.．.“SigMean”；“SigMedian”；“SigRMS”；“SigVar”；“SigPeak”；“SigPeak2Peak”；“SigSkewness”；.．.“SigKurtosis”；“SigCrestFactor”；“SigMAD”；“SigRangeCumSum”；“SigCorrDimension”；“SigApproxEntropy”；.．.“SigLyapExponent”；“PeakFreq”；“HighFreqPower”；“EnvPower”；“PeakSpecKurtosis”];如果runLocal而Hasdata (ens) data = read(ens);addData = analyzeData(数据);writeToLastMemberRead(实体,addData);结束其他的将集成拆分为多个分区，并并行分析每个分区。N = numpartitions(ens,gcp);parforCt = 1:n subens =分区(ens,n, Ct);而Hasdata (subens) data = read(subens);addData = analyzeData(数据);writeToLastMemberRead (suben, addData)结束结束结束

featureVariables = analyzeData(“GetFeatureNames”）;ens.DataVariables = [ens.DataVariables;featureVariables];ens.SelectedVariables = [featureVariables;ens.ConditionVariables];重置(实体)

featureData = gather(tall(ens))

idxResponse = strcmp(featureData.Properties.VariableNames，“SensorDrift”）;idxLastFeature = find(idxResponse)-1;用作潜在预测因子的最后一个特征的指数featureAnalysis = fscnca(featureData{:，1:idxLastFeature}，featureData. sensordrift);featureAnalysis。FeatureWeights

idxSelectedFeature = featureAnalysis。FeatureWeights > 0.1;classifySD = [featureData(:，idxSelectedFeature)， featureData(:，idxResponse)]

图直方图(classifySD.SigRangeCumSum (classifySD.SensorDrift),“BinWidth”5 e3)包含(“信号累积和范围”) ylabel (“数”)举行在直方图(classifySD.SigRangeCumSum (~ classifySD.SensorDrift),“BinWidth”、5 e3)从传奇(“传感器漂移故障”，“无传感器漂移故障”）

idxResponse = strcmp(featureData.Properties.VariableNames，“ShaftWear”）;featureAnalysis = fscnca(featureData{:，1:idxLastFeature}，featureData. shaftwear);featureAnalysis。FeatureWeights

idxSelectedFeature = featureAnalysis。FeatureWeights > 0.1;classifySW = [featureData(:，idxSelectedFeature)， featureData(:，idxResponse)]

图直方图(classifySW.SigLyapExponent(classifySW.ShaftWear))信号李雅普诺夫指数) ylabel (“数”)举行在直方图(classifySW.SigLyapExponent (~ classifySW.ShaftWear))从传奇(“轴磨损故障”，“无轴磨损故障”）

idxResponse = strcmp(featureData.Properties.VariableNames，“ToothFault”）;featureAnalysis = fscnca(featureData{:，1:idxLastFeature}，featureData. toothfault);[~，idxSelectedFeature] = sort(featureAnalysis.FeatureWeights);classifyTF = [featureData(:，idxSelectedFeature(end-5:end))， featureData(:，idxResponse)]

图直方图(classifyTF.SigRangeCumSum(classifyTF.ToothFault))“信号累积和范围”) ylabel (“数”)举行在直方图(classifyTF.SigRangeCumSum (~ classifyTF.ToothFault))从传奇(“齿轮齿故障”，“无齿轮齿故障”）

rng (“默认”）%用于再现性cvp = cvpartition(size(classifyTF,1)，“KFold”5);%随机划分数据进行训练和验证分类tf = fitcsvm(.．.classifyTF (cvp.training (1), 1: end-1),.．.classifyTF (cvp.training(1)、结束),.．.“KernelFunction”，多项式的，.．.“PolynomialOrder”2,.．.“KernelScale”，“汽车”，.．.“BoxConstraint”, 1.．.“标准化”,真的,.．.“类名”,(假;真正的]);

使用测试验证数据上的分类器来评估性能actualValue = classifyTF{cvp.test(1)，end};predictedValue = predict(classifierTF, classifyTF(cvp.test(1)，1:end-1));通过计算和绘制混淆矩阵来检查性能confdata = confusimat (actualValue,predictedValue);H =热图(confdata，.．.“YLabel”，“实际齿轮齿故障”，.．.“YDisplayLabels”, {“假”，“真正的”}，.．.“包含”，“预测齿轮齿故障”，.．.“XDisplayLabels”, {“假”，“真正的”}，.．.“ColorbarVisible”，“关闭”）;