基于SA-PSO算法优化CNN的电能质量扰动分类模型

针对传统电能质量扰动分类模型中扰动特征复杂、识别步骤繁琐的问题，提出了一种通过模拟退火(SA)算法与粒子群优化(PSO)算法相结合来优化卷积神经网络(CNN)的电能质量扰动分类模型。将CNN卷积层中的二维卷积核替换成一维卷积核；采用SA算法对PSO算法进行改进，规避PSO算法陷入局部最优的困境；采用改进后的PSO算法对CNN进行参数寻优；利用优化CNN提取和筛选合适的特征，根据这些特征利用分类器得到最终分类结果。通过算例分析得出，使用基于SA-PSO算法优化的CNN的电能质量扰动分类模型能精确地识别出电能质量扰动信号。

Power quality disturbance classification model based on CNN optimized by SA-PSO algorithm

Aiming at the problems of complex disturbance features and complicated recognition steps in traditional power quality disturbance classification models, a power quality disturbance classification model based on convolutional neural network(CNN) optimized by combining simulated annealing(SA) algorithm and particle swarm optimization(PSO) algorithm is proposed. The two-dimensional convolution kernel in the CNN convolution layer is replaced by one-dimensional convolution kernel. The SA algorithm is used to improve the PSO algorithm to avoid the PSO algorithm falling into the local optimal dilemma. Then, the improved PSO algorithm is used to optimize the parameters of CNN. The improved CNN is used to extract and screen appropriate features, according to which, the final classification results are obtained by the classifier. Through the example analysis, it is concluded that the power quality disturbance classification model based on CNN optimized by SA-PSO algorithm can accurately identify the power quality disturbance signal.