摘要
谐波的检测分析对于电力系统谐波治理非常重要,但常用的FFT检测方法存在频谱泄漏及栅栏现象等缺陷,加窗插值算法的使用一定程度上弥补了这些不足却又增大了计算量和存储容量要求;人工神经网络具有快速处理数字信号能力,本文以谐波离散傅里叶变换后的三角函数和傅里叶系数分别作为BP网络的隐层神经元和权值可获得了一种训练速度更快的神经网络;通过该神经网络算法和效果相对较好的几种FFT插值算法的仿真实例比较,验证了该算法能够更快更精确地对电力系统谐波进行分析,对谐波治理具有较大意义。
Harmonic detection and analysis is very important measures to harmonic elimination in power system. Some disadvantages such as spectrum leakage and barrage phenomenon exist in the common FFT detection method, and the windowed interpolation algorithm can make up some shortcomings at a certain degree but increases the demand of computational complexity and storage capacity. The artificial neural network has high-speed processing capability. This paper uses the trigonometric function after discrete fourier transform and the Fourier coefficients as BP network hidden neurons and weight respectively, a faster training speed neural network can be obtained. By the simulation comparison based on the neural network algorithm and several FFT detection methods with relatively good effects, it is be verified that this neural network algorithm can analyze the power system harmonic more fastly and precisely which has more important significance on harmonic eliminafinn
出处
《电瓷避雷器》
CAS
北大核心
2014年第6期67-71,共5页
Insulators and Surge Arresters
关键词
谐波分析
傅里叶变换
BP神经网络
harmonic analysis
Fourier transform
BP neural network
