基于PCA-PNN的柴油机故障识别方法研究

Research on Fault Identification Method of Diesel Engine Based on PCA-PNN

利用Matlab/Simulink搭建柴油机仿真模型对柴油机的热工故障进行仿真预测,获得柴油机热工故障参数,使用主成分分析法(PCA)计算柴油机热工故障参数的综合统计量,通过综合统计量的偏移程度检测柴油机是否发生故障,结果表明:PCA可以准确地检测出柴油机是否发生故障.并使用主成分分析法对柴油机热工故障参数进行分析计算,选取能够反映原始数据97.26%信息的三个主成分作为概率神经网络(PNN)的输入,将柴油机的故障模式作为输出,构建一个四层的概率神经网络预测模型,结果表明:PCA-PNN模型能够很好地对柴油机的故障模式做出诊断.

Matlab/Simulink was used to build a diesel engine simulation model to simulate and predict the thermal fault of diesel engine,so as to obtain the thermal fault parameters of diesel engine.The principal component analysis(PCA)was adopted to calculate the comprehensive statistics of thermal fault parameters of diesel engine,and the deviation degree of the comprehensive statistics was used to detect whether the diesel engine fails.The results show that PCA can accurately detect whether the diesel engine fails.Then,the principal component analysis method was used to analyze and calculate the thermal fault parameters of diesel engine,and three principal components which can reflect 97.26%information of the original data were selected as the inputs of probabilistic neural network(PNN).Taking the failure mode of diesel engine as the output,a four-layer probabilistic neural network prediction model was constructed,and the results show that PCA-PNN model can well diagnose the failure mode of diesel engine.