基于改进PSO和DE优化神经网络的电能质量扰动分类  

Power quality disturbance classification based on improved PSO and DE optimized neural network

作　　者：金梅 张伟亚 张淑清 张立国 颜庭鑫 Jin Mei;Zhang Weiya;Zhang Shuqing;Zhang Liguo;Yan Tingxin(Key Laboratory of Measurement Technology and Instrumentation of Hebei Province,Qinhuangdao 06600)

机构地区：[1]河北省测试计量技术与仪器重点实验室,秦皇岛066004

出　　处：《高技术通讯》2018年第4期291-298,共8页Chinese High Technology Letters

基　　金：国家自然科学基金(61077071);河北省自然科学基金(F2016203496;F2015203413)资助项目

摘　　要：针对电能质量问题提出了基于改进粒子群优化算法(PSO)和差分进化算法(DE)相结合优化神经网络的分类方法。首先用Matlab仿真几种典型的电能质量扰动信号,再利用小波变换进行多尺度的分解,得到各尺度的能量信息作为特征向量输入BP神经网络分类器中对扰动信号进行快速、准确的分类识别。并针对传统BP算法收敛耗时长速度慢,不能保证获得全局最优等缺点,在种群分类基础上提出了一种混合粒子群与差分进化算法的新型PSO-DE算法,并利用其对神经网络进行改进。这种混合PSO-DE算法在很大程度上能弥补BP神经网络的不足,采用该算法对网络进行优化后完成电能质量扰动信号的自动分类。Aiming at the power quality problem,an improved classification method based on improved particle swarm optimization（ PSO） and differential evolution algorithm（ DE） optimization neural network is proposed. Firstly,several typical power quality disturbances are simulated by Matlab,and multi-scale decomposition of wavelet transform is used to obtain the energy features of each scale,which is used as input feature vectors of neural networks. Then BP network is used to train the output samples,and the input feature vectors are identified and classified. In view of the shortcomings of the traditional BP algorithm,such as easiness to fall into local minimum and slow convergence speed,a new PSO-DE hybrid algorithm is introduced to improve the neural network. The algorithm is based on improved particle swarm optimization and differential evolution algorithm for population classification. This hybrid PSODE algorithm can make up for the deficiency of BP neural network to a great extent. After the optimization of the network,the algorithm can automatically classify the power quality.

关 键 词：改进粒子群优化算法(PSO) 差分进化算法(DE) 神经网络 电能质量 扰动分类 

分 类 号：TM76[电气工程—电力系统及自动化] TP18[自动化与计算机技术—控制理论与控制工程]