基于Elman神经网络与PSO算法的机械密封端面摩擦状态识别

摘要

Acoustic emission method can be used to monitor the friction state of the mechanical seal face.In order to ex-tract the features of acoustic emission signals accurately,a method that used particle swarm optimization ( PSO) algorithm to optimize Elman neural network was presented.The friction state of mechanical seal face was identified by this method, and the recognition rate the Elman neural network before and after optimizing by PSO algorithm was compared.The results show that the Elman neural network optimized by PSO algorithm has higher recognition rate of the friction state of mechani-cal seal face,so the effective monitoring of the friction state of mechanical seal face in real time is realized.%声发射法可用于监测机械密封工作过程中端面的摩擦状态。为准确提取机械密封端面声发射信号特征，提出了一种利用PSO算法（Particle Swarm Optimization，粒子群优化）对Elman神经网络进行优化的方法。采用该方法对机械密封端面的摩擦状态进行识别，并比较了优化前后神经网络对机械密封端面摩擦状态的识别率。结果表明：经过PSO算法优化后的 Elman神经网络对机械密封的端面摩擦状态有更高的识别率，从而实现了对机械密封端面摩擦状态实时有效的监测。