基于小波神经网络的电缆局部放电自动诊断方法研究

Research on Automatic Diagnosis Method of Cable Partial Discharge Based on Wavelet Neural Network

为科学高效地诊断电缆绝缘状态,保障城区配电网的安全稳定运行,提出一种改进鲸鱼算法与小波神经网络结合的电缆局部放电诊断方法。首先利用灰度图谱与矩特征实现对局部放电信号的特征提取,基于矩特征值构建小波神经网络的输入样本集;然后使用改进的鲸鱼优化算法对小波神经网络参数进行寻优,以解决神经网络普遍存在的参数敏感问题;最后将优化好且训练完成的小波神经网络应用于电缆局部放电诊断中。改进的鲸鱼优化算法引入非线性收敛因子与自适应思想提升了算法的性能,对小波神经网络的超参数有较好的寻优效果。仿真结果表明,所提出的方法能够有效地诊断电缆绝缘缺陷类型,诊断精度相比于通用参数配置的小波神经网络有9.45%的提升。

In order to diagnose the cable insulation state scientifically and efficiently and ensure the safe and stable operation of the urban distribution network,a cable partial discharge diagnosis method based on the combination of improved whale algorithm and wavelet neural network was proposed;firstly,use the gray-scale atlas and moment features to extract the features of the partial discharge signal,and construct the input sample set of the wavelet neural network based on the moment eigenvalues;then,the adaptive whale optimization algorithm was used to optimize the parameters of wavelet neural network to solve the parameter sensitivity problem commonly existed in the neural network;finally,the optimized and trained wavelet neural network was applied to the cable partial discharge diagnosis.The improved whale optimization algorithm introduces nonlinear convergence factor and self-adaptive idea to improve the performance of the algorithm,and has a good optimization effect on the hyper parameters of the wavelet neural network.The simulation results show that the proposed method can effectively diagnose the types of cable insulation defects,and the diagnosis accuracy is improved by 9.45% compared with the wavelet neural network with general parameter configuration.