基于改进型神经网络的电能质量扰动识别

为了改善神经网络的电能质量扰动识别能力,提出了一种改进型神经网络。在分析了传统BP神经网络和遗传算法优化BP神经网络(GA-BP)的基础上,将遗传算法和贝叶斯正则化神经网络相结合,并采用小波包能量熵作为特征向量。改进后的神经网络能有效克服传统BP神经网络易陷入局部最小,GA-BP易出现过拟合现象且网络节点数偏多等缺点。在MATLAB平台上建立各种电能质量扰动信号的仿真模型,分别采用传统BP神经网络、GA-BP及改进型神经网络进行扰动识别对比。仿真结果表明,改进后的神经网络显著提高了识别正确率。

Power Quality Disturbance Identification Based on Improved Neural Network

An improved neural network is put forward in this paper to enhance the identification effect of power quality disturbance. Based on analyzing the traditional BP neural network and the genetic algorithm optimized BP neural network{ GA-BP）, a new method based On Bayesian-regularization neural network is presented, and the wavelet packet-energy entropy is used to construct a feature vector. The improved method can overcome the local minimum problem of traditional BP neural network, the overfitting and too much network nodes of GA-BP. The simulated models of the disturbance signals were built, and tested by traditional BP neural network, GA-BP and the improved neural network, respectively. The simulation results show that the improved neural network can enhance the identification accuracy significantly.