基于HPSO-WNN的牵引变压器故障诊断算法研究

摘要

为全面有效地诊断电力机车牵引变压器故障,提出一种基于混合粒子群算法的正交小波神经网络（HPrnSO-WNN)方法,对牵引变压器进行综合测试和诊断.将色谱数据和电气试验数据作为正交小波神经网络的输rn入量,网络隐藏层采用具有正交性的小波函数db4作为基函数,利用混合粒子群算法获得正交小波神经网络的初rn始值并优化网络参数.试验结果证明,本文提出的HPSO-WNN确实可有效提高牵引变压器故障诊断速度和准rn确度.%In order to diagnose traction transformer faults more effectively and in an all-round way, this paper proposed the method of the orthogonal wavelet neural network based on the hybrid particle swarm optimization algorithm (HPSO-WNN) to be applied in comprehensive tests and diagnoses of electric locomotive traction transformer faults. The chromatographic data and electrical test data worked as the inputs of the orthogonal wavelet neural network, the network's hidden layer used the orthogonal dh4 function as the basis function, and the hybrid particle swarm algorithm was used to obtain the initial values of the orthogonal wavelet neural network and to optimize the network parameters. The test results show that the proposed HPSO-WNN does effectively improve the traction transformer fault diagnosis speed and accuracy.