开关柜局部放电模式识别

Partial discharge pattern recognition of switchgear

针对高压开关柜局部放电模式分类中样本数较少,常规的分类方法识别率较低,提出了一种基于改进果蝇算法(FOA)优化的概率神经网络模型(PNN)的局部放电模式识别方法。作为一种新型的群体搜索随机优化算法,FOA算法具有原理简单、参数少,收敛快等优点。实验采集的局部放电信号进行处理并归一化,作为FOA-PNN神经网络的输入样本进行识别,并与传统算法进行比较。仿真实验证明:该算法可以有效地运用到局部放电模式识别中,为开关柜局部放电诊断供了一条新途径,具有良好的研究价值和发展前景。

In view of the small number of samples and the low recognition rate of conventional classification methods in high-voltage switchgear partial discharge pattern classification,this paper proposes a method of partial discharge pattern recognition based on improved Drosophila algorithm(FOA)optimized probabilistic neural network(PNN)model.As a new group search stochastic optimization algorithm,FOA algorithm has the advantages of simple principle,few parameters and fast convergence.The experimental partial discharge signals are processed and normalized,and identified as input samples of FOA-PNN neural network,and compared with traditional algorithms.The simulation results show that the algorithm can be effectively applied to partial discharge pattern recognition,which provides a new way for the diagnosis of partial discharge in switchgear,and has good research value and development prospects.