基于广义规则归纳算法的电力稽查关联问题分析模型研究

电力稽查是电力营销中不可或缺的一个职能,在电力营销市场化过程中扮演越来越重要的角色。如何平衡稽查与被稽查者之间的利益,在有限的稽查中发现和预测电力营销中更多的错误,已成为电力公司和科研人员关注的新课题。本文从建模角度出发,对稽查样本问题库进行编码整理,再对问题字段进行bool变换,将文本规范化为符号变量,解决了文本难以分类和噪声过大的问题。然后利用数据挖掘中的GRI关联算法对已有的稽查样本数据库进行问题关联分析,得出问题与问题间的隐含关系。并使用广州市近一年来的稽查数据样本进行实例分析,推算出100条可信的关联规则,验证了模型的可靠性和有效性。

The study of GRI model for association problem of electrical inspection

The inspection is an indispensable function in power marketing, and which plays an increasing important role in the process of power marketing. How to balance the interests of inspector and inspected man, discover or forecast more and more mistake in the process of power audit has become a new topic in the electric power and relevant researcher. By pretreated and coded the inspected sample problem, we transformed the problem to the structured data and these disposed data could be run well by mathematical models, which solved the difficult of text classify and big noise. Then, by used the GRI （ Generalized Rule Induction） algorithm, we studied and analysed the association of problem in sample data, and acquired the implied relationship between different problem. In this paper, our research showed that the GRI model was effective and useful for the analysis of association problem.