基于PSO-PNN与CV-SVM的旋转机械故障诊断研究

Research of rotating machinery fault diagnosis based on PSO-PNN and CV-SVM

不同类型的旋转机械发生故障时会激发出不同特征的振动信号。针对旋转机械故障点位判断难、复合故障判断不准确等问题,构建了概率神经网络(PNN)以及支持向量机(SVM)这两种人工智能模型,并采用该模型对旋转机械进行了故障识别研究。首先,采集了研究对象各故障状态下的振动信号,对振动信号的时域和频谱进行了分析,根据振动信号的特征表现,分别将原始振动信号幅值和振动信号特征值作为人工智能模型的输入向量;然后,利用粒子群算法(PSO)对概率神经网络的输入参数进行了优化,利用交叉验证法(CV)对支持向量机的输入参数进行了优化;最后,建立了概率神经网络和支持向量机故障诊断模型,对旋转机械故障进行了诊断,并对比分析了诊断结果。研究结果表明:基于PSO-PNN模型的旋转机械故障识别准确率在97%以上;基于CV-SVM模型的旋转机械故障识别准确率在98%以上;这两种人工智能方法在用于旋转机械故障诊断时具有速度快、准确率高的优点;其中,PSO-PNN方法适用于旋转机械故障实时监测,CV-SVM方法适用于旋转机械复杂故障的识别。

Different types of rotating machinery faults excited vibration signals with different characteristics.Aiming at the problem that it was difficult to judge the fault point of rotating machinery and the compound fault was not accurate,two kinds of artificial intelligence models,probabilistic neural network(PNN)and support vector machine(SVM)were constructed to identify rotating machinery faults.Firstly,the vibration signals of the research object under each fault state were collected,and the time domain and frequency spectrum of the vibration signals were analyzed.According to the characteristic performance of the vibration signals,the original vibration signal amplitude and vibration signal characteristic value were respectively used as the input vector of the artificial intelligence model.Then,the particle swarm optimization(PSO)was used to optimize the input parameters of the probabilistic neural network,and the cross validation(CV)was used to optimize the input parameters of the support vector machine.Finally,the probabilistic neural network and support vector machine fault diagnosis model were established to diagnose rotating machinery faults,and the diagnosis results were compared and analyzed.The results show that the accuracy of rotating machinery fault identification based on PSO-PNN model is more than 97%.The accuracy of fault identification of rotating machinery based on CV-SVM model is more than 98%.These two artificial intelligence methods have the advantages of fast speed and high accuracy in the fault diagnosis of rotating machinery.PSO-PNN method is suitable for real-time monitoring of rotating machinery faults,and CV-SVM method is suitable for identifying complex rotating machinery faults.