基于粒子群与神经网络的间谐波测量算法研究  

Hybrid algorithm for interharmonics measurement based on particle swarm optimization and neural network

作　　者：覃思师[1] 刘前进[1] 

机构地区：[1]华南理工大学电力学院,广东广州510640

出　　处：《电力系统保护与控制》2011年第2期18-22,共5页Power System Protection and Control

摘　　要：为了精确测量电力系统的非整数次谐波,提出一种基于粒子群与神经网络的混合算法。该算法通过FFT变换得出谐波个数和精度不高的谐波幅值、相位、谐波次数,然后初始化粒子群,再由粒子群优化算法训练神经网络,得出间谐波的各项参数。同时提出一种基于可变参数的神经元激发函数,使得谐波次数和权值一样参与调整,更有利于检测非整数次谐波。仿真实例表明,该算法能将频率相近的非整数次谐波分离,可快速、精确地获得非整数次谐波的各项参数。In order to measure the power system interharmonics accurately,this paper presents a hybrid algorithm for interharmonics measurement based on particle swarm optimization and neural network.The sampled signal is processed with FFT algorithm,then its number,magnitudes,phases,and orders of harmonics are obtained.After initializing the particle swarm and training the neural network by the particle swarm optimization algorithm,the parameters of interharmonics can be gotten.At the same time,the paper proposes the neural excitation function based on variable parameters,which makes the number of harmonics and weights of neural network participate in the adjustment,thereby it is better for detecting the non-integer harmonics.Simulation results show that close non-interger harmonics can be separated from a signal and parameters of interharmonics are obtained fast and accurately by the algorithm.

关 键 词：电力系统 神经网络 快速傅里叶变换 粒子群优化 谐波分析 

分 类 号：TM711[电气工程—电力系统及自动化]