基于近邻传播算法和随机森林回归模型的台区线损率计算方法

Calculation Method for Line Loss Rate in Transformer District Based on Affinity Propagation Algorithm and Random Forest Regression Model

针对低压配网接线方式复杂和线损率难以准确计算的问题,本文提出一种基于近邻传播聚类算法和随机森林回归模型的台区线损率计算方法。基于线路损耗模型提出了台区线损率预测计算的电气特征指标,并利用主成分分析方法提取适用于聚类分析的主特征参数,然后采用近邻传播聚类算法对数据进行聚类分析。在此基础上,采用随机森林回归算法对每类聚类数据进行样本的训练学习,并利用包外数据进行预测。以某地区614个台区样本进行仿真计算,仿真结果验证了本文所提算法的有效性和正确性,并且计算精度要优于多元线性回归算法。

In view of the complex connection mode of low-voltage distribution network,it is difficult to accurately calculate the line loss rate.To solve this problem,a calculation method for the line loss rate in a transformer district is put forward in this paper,which combinesthe affinity propagation(AP)clustering algorithm with the random forest regression(RFR)model.Based on the line loss model,the electrical characteristic indexes for the calculation of line loss rate in the transformer district are proposed,and the main characteristic parameters suitable for clustering analysis are extracted usingthe principal component analysis method.Then,the data are clustered by the AP clustering algorithm.On this basis,the RFR algorithm is used to train the samples of each cluster and predict with the out-of-bag data.During simulations,614 samples of transformer district are calculated,and simulation results verify the validity and correctness of the proposed method,with acalculation accuracy better than that of the multiple linear regression(MLR)algorithm.