粒子群优化的小波算法在避雷器老化诊断中的应用

摘要

为研究金属氧化物避雷器漏电流"消噪"问题,提出一种粒子群优化的小波消噪方法.首先利用db5小波对漏电流信号进行5层分解,其次设定待求解阈值,对信号进行重构,最后通过粒子群优化求解阈值实现消噪,并通过模拟MOA小电流区模型进行仿真验证.研究表明:使用db5小波对信号进行分解,并利用PSO对阈值进行优化求解,最终阈值c5,c4,c3,c2,c1分别为0.32,0.20,0.13,0.02和0.01.消噪后信噪比相对于单独使用平稳小波提升了7 dB,说明利用结合粒子群优化的小波消噪算法进行消噪,消噪效果明显优于单独使用小波消噪算法.%To solve the de-noise problem of metal oxide arrester (MOA) leakage current, a PSO (particle swarm optimization)-based wavelet de-noising algorithm is proposed. Firstly, the db5 wavelet is used to decompose the leakage current. Secondly, the threshold value is set and the processed wavelet coefficients are reconstructed. Finally, de-noising is achieved through PSO threshold value and the results are verified through MOA current modelling. Studies show that by decomposing the leakage current with db5 and optimizing the threshold value with PSO, the values ofc5,c4,c3,c2 andc1 are found to be 0.32, 0.20, 0.13, 0.02 and 0.01, respectively. The SNR (signal to noise ratio) after de-noise is raised by 7dB compared with the result when just using the stationary wavelet. The results indicate that the de-noising effect of PSO-based wavelet de-noising algorithm is better than that of wavelet de-noising algorithm.