基于SVM修正模糊多属性决策法的用电客户信用评价

针对电力客户信用评价缺乏准确的科学指标体系和评价中主观因素和经验因素难以量化、标准化的问题,提出了支持向量机(SVM)修正的模糊期望值决策法。建立了电力客户信用评价指标体系,并以三角模糊数的形式给出指标值和评价者的主观经验值,利用模糊期望值决策法得到信用期望值,最后通过SVM基于结构风险最小化原则修正该期望值。实例分析表明,SVM修正后的电力客户信用等级与供电企业原先为其设定的信用等级基本符合,更可靠地贴近实际。

Application of fuzzy decision making method based on expected value modified by SVM in the evaluation of electricity customers' credit

Fuzzy decision making method based on expected value modified by SVM was proposed in view of evaluation of electricity customers' credit lacking of precise index system and in the evaluation, subjective factors and experience factors that are hard for quantity and standardization. First index system for evaluation of electricity customers' credit is established. The index and the subjective experience value of valuators are given in the form of triangular fuzzy number. Then expected values were resulted by fuzzy decision making method based on expected value. Finally, structure risk minimization principle modifies the expected values. The experiment shows that the credit grades by modification suit to the original credit grades enacted by power supply enterprise and are more practicable.