摘要
我国电力市场化改革进程不断推进,将作为市场主体的电力用户信用评价问题分析清楚对电力市场的构建具有重要而深远的意义。针对电力用户信用评价问题提出一种基于层次分析与Adaboosting的电力用户信用综合评价方法。基于电力用户数据信息,建立了归一化综合评价指标体系;构建了基于层次分析法的信用评价模型,从典型分类算法中遴选出合适的备选分类算法;采用Adaboosting算法对备选分类算法进行线性动态赋权,通过对样本数据训练建立组合分类模型。基于实际电力用户信息数据对提出的综合评价方法进行应用验证,结果表明:相比已有的分类算法,所提出的综合评价方法的评价准确率与曲线下面积(AUC)值均较高,具有较好的适用性。
With the reform of the electric marketization,it is of great and far-reaching significance for the construction of the electricity market to analyze clearly the credit evaluation problems of power users as the main market. Aiming at the problem of power user credit evaluation,a comprehensive evaluation method based on the analytic hierarchy process(AHP)and Adaboosting algorithm was proposed. Based on the power user data information,a normalized comprehensive evaluation index system was established. A credit evaluation model based on the AHP was constructed,and the suitable candidate classification algorithms were selected out from the typical classification algorithms. The Adaboosting algorithm was adopted to perform the linear dynamic weighting on the candidate classification algorithms,and a combined classification model was established by training sample data. The proposed method was verified with the actual power user information data. The results show that the proposed method has higher evaluation accuracy and area under curve(AUC)value,and has better applicability with the existing classification methods.
作者
徐宏宽
林顺富
边晓燕
李东东
XU Hongkuan;LIN Shunfu;BIAN Xiaoyan;LI Dongdong(College of Electrical Engineering,Shanghai University of Electric Power,Shanghai 200090,China)
出处
《电气传动》
2023年第2期79-85,共7页
Electric Drive
基金
国家自然科学基金(51977127)
上海市科学技术委员会资助项目(19020500800)
上海市教育发展基金会和上海市教育委员会“曙光计划”资助。
