基于聚类算法的配电网线损预测方法

Line Loss Prediction Method of Distribution Network Based on Clustering Algorithm

线损预测可以为配电网线损水平评估提供一种有效途径。为提高配电网线损预测精度,提出一种基于聚类算法的配电网线损预测方法。利用线损数据的空间比值,计算出冗余度指数,判断配电网线损数据的冗余程度,利用聚类算法聚类处理线损数据。利用瞬态自适应麻雀搜索算法(Transient Adaptive Sparrow Search Algorithm,TASSA)优化多粒度长短期记忆(Multi granularity-Long Short Term Memory,Mg-LSTM)网络,剔除异常的线损数据,通过训练优化后的Mg-LSTM网络,构建线损预测模型,从而预测配电网的线损。实验结果表明,所提方法可以将线损预测误差率控制在2%以内,提高了配电网线损预测精度。

Line loss prediction can provide an effective way to evaluate the line loss level of distribution networks.In order to improve the accuracy of line loss prediction for distribution networks,a line loss prediction method for distribution networks based on clustering algorithm is proposed.Using the spatial ratio of the line loss data,the redundancy index is calculated to determine the redundancy level of the distribution network line loss data,and the clustering algorithm is used to cluster the line loss data.Transient Adaptive Sparrow Search Algorithm(TASSA)is used to optimize the Multi granularity-Long Short Term Memory(Mg-LSTM)network to eliminate abnormal line loss data.By training the optimized Mg-LSTM network,a line loss prediction model is constructed to predict the line loss of the distribution network.The experimental results show that the method in the paper can control the error rate of line loss prediction within 2%,which improves the accuracy of line loss prediction in distribution networks.