基于特征选择的IDBO-ELM配电网台区线损计算

Line Loss Calculation of the IDBO-ELM Distribution Network Transformer Area Based on Feature Selection

中文关键词:配电网  线损计算  特征选择  极限学习机

keywords:distribution network  power loss calculation  feature selection  exteme learning machine

中文摘要:低压配电台区用户数目庞大、线路复杂，传统线损计算工作量大，新型负荷接入配电台区增加了线损计算难度。该文提出改进蜣螂算法（improved dung beetle optimizer，IDBO）优化的极限学习机（exteme learning machine，ELM）配电台区线损计算方法，综合分析多元因素与台区线损关联性，得到影响台区线损计算准确度的特征因素；将台区线损数据和影响因素进行归一化处理形成数据集，采用交叉验证法划分数据集得到训练集和测试集；采用网格搜索法确定ELM隐藏层节点数，并用IDBO确定ELM隐藏层权值和偏差，实现对ELM算法的优化，建立IDBO-ELM配电台区线损计算模型。通过与现有模型进行算例对比分析可得所提出的模型线损计算均方根误差为0.25，平均绝对百分比误差为0.63%，决定系数可达0.98，能够实现对配电网台区线损的准确计算。

Abstract:The low-voltage distribution transformer area is characterized by a large number of users and complex line arrangements， which poses challenges for traditional line loss calculation due to its time-consuming nature. The advent of new types of load has further complicated the process. To address these issues， this paper proposes a distribution transformer area line loss calculation model based on an optimized extreme learning machine （ELM） enhanced by an improved dung beetle optimizer （IDBO）. This method involves a comprehensive analysis of the correlation between various factors and substation line loss to identify key factors that impact calculation accuracy. The data on line losses and influencing factors are normalized and processed into a dataset， which is then divided into training and test sets using cross-validation. The optimal number of hidden layer nodes for ELM is determined through grid search， while the hidden layer weight and bias are determined by IDBO， thus optimizing the ELM algorithm and establishing the IDBO-ELM model for the line loss calculation of the distribution transformer area. Through comparative analysis with existing models， it is found that the root mean square error of the line loss calculation of the proposed model is 0.25， the mean absolute percentage error is 0.63%， and the coefficient of determination is up to 0.98， indicating that it can achieve accurate calculation of line loss in the distribution transformer area.