基于概率神经网络的TSI发动机故障识别方法

Research on the recognizing method of TSI engine trouble based on probabilistic neural net

针对TSI发动机故障检测问题,应用概率神经网络技术,分析了执行器工作波形上的幅值、周期、波形走势和脉宽等特征值,诊断故障原因。同时,将提取的特征值利用MATLAB软件进行数据处理与分组,建立PNN神经网络并进行训练与测试。以喷油器驱动器的工作电压波形为例,对所设计的故障诊断方法进行验证,并将PNN神经网络与传统BP神经网络的故障诊断结果准确率进行对比。实验结果表明:基于概率神经网络的故障识别方法可快速准确地识别喷油器故障。

In view of the TSI engine fault detection problem,the characteristics of the amplitude,cycle,waveform,pulse width and pulse width of the actuator working waveform are analyzed by using probabilistic neural network technology. At the same time, the extracted eigenvalues are processed and grouped using MATLAB software, and the PNN neural network is es- tablished and trained and tested. Taking the working voltage waveform of injector actuator as an example, the designed fault diagnosis method is validated, and the accuracy of the fault diagnosis result of PNN neural network is compared with that of the traditional BP neural network. The experimental results show that the fault identification method based on probabilistic neural network can identify injector faults quickly and accurately.