面向新型电力系统的配网供电数据分析与处理技术

Data Analysis and Processing Technology of Distribution Network Power Supply for New Power System

针对现有技术方案对新型电力系统中配网电能质量识别准确率较低的问题,提出了一种基于经验模态分解(empirical mode decomposition,EMD)、灰狼优化(grey wolf optimizer,GWO)算法和支持向量机(support vector machine,SVM)算法的电能质量识别分类融合模型。该模型以配电网监测的电压及电流波形作为输入,通过EMD算法对输入数据进行特征提取,得到多个固有模态分量。同时,利用GWO算法对SVM模型的惩罚系数与核函数参数进行优化,并将固有模态分量作为最优SVM模型的输入特征,从而实现电能质量问题的精准识别分类。算例分析结果表明,所提子模块的改进算法能够实现波形特征的分解提取以及参数的快速优化,且EMD-GWO-SVM融合算法的电能质量分类识别平均准确率可达92.83%。

In view of the low accuracy of the existing technical schemes in the identification of power quality in the distribution network in the new power system,a classification fusion model of power quality identification based on empirical mode decomposition(EMD),gray wolf optimizer(GWO)algorithm and support vector machine(SVM)algorithm was proposed.The model took the voltage and current waveforms monitored by the distribution network as the input,and extracted the characteristics of the input data through EMD algorithm to obtain the components of multiple natural modes.At the same time,GWO algorithm was used to optimize the penalty coefficient and kernel function parameters of the SVM model,and the natural mode component was used as the input feature of the optimal SVM model,so as to achieve accurate identification and classification of power quality problems.The results of the example analysis show that the improved algorithm of the proposed sub-module can achieve the decomposition and extraction of the waveform features and the rapid optimization of the parameters,and the average accuracy of the EMD-GWO-SVM fusion algorithm for power quality classification and recognition can reach 92.83%.