基于PSO神经网络的定子电流信号谐波检测

Study on Algorithm of Stator Current Signal Harmonic Detection Based on PSO Neural Networkl

定子电流信号检测法是电机故障诊断常用方法之一,精确的谐波、间谐波参数检测是其前提条件。为了提高电机定子电流信号谐波、间谐波检测精度,提出一种电机定子电流信号谐波、间谐波检测PSO神经网络算法。将电机定子电流采样信号用加窗FFT算法预处理,获得谐波、间谐波的个数和精度较低的谐波、间谐波次数、幅值和相位。将这些参数作为粒子群初始化的依据,用PSO算法训练神经网络,最终得到高精度的谐波、间谐波各项参数。实例仿真表明,该算法与加窗FFT相比,能快速、精确地检测电机定子电流信号谐波、间谐波各项参数,为准确的诊断电机故障奠定了基础。

The stator current signal testing is one of common methods of the motor fault diagnosis.Measurement of accurate harmonics and inter-harmonics parameters is the basement.In order toimprove detecting precision,a PSO neural network algorithm is presented for detecting the harmonicsand inter-harmonics of the motor stator current signal.The sample signal of motor stator current ispreprocessed with windowed FFT.Then the number and low precision of magnitudes,phases,and ordersof harmonics and inter-harmonics can be gained.These parameters are initialized as particles swarmoptimization input.After training the neural network by the particle swarm optimization algorithm,theaccurate parameters of harmonics and inter-harmonics can be gotten.Simulation results show that thisone can obtain harmonics and inter-harmonics parameters of the motor stator current signal fast andaccurately,compared with windowed FFT algorithm.This work attributes important basement for thefault diagnosis of motor.