基于一种新的启发式约简算法的变压器故障诊断方法

基于依赖度的知识相对约简的启发式约简算法是一种新的粗糙集最优属性约简方法,本文首次将其应用于变压器故障诊断问题中,并结合粗糙集值约简方法,得到一组故障诊断的最小决策规则集,从而大大减小了编码的工作量,避免了约简属性组合查询及缺少关键属性时规则匹配所带来的不便,运算速度也相对加快。此外,该模型还可以通过丰富训练样本,修正决策表的自学习法使得诊断效果不断提高。最后结合实例分析,证明该方法的简便及有效。

Diagnosis model of transformer faults based on a new heuristic reduction algorithm

Heuristic reduction algorithm which is proposed on the dependency of a knowledge decision system is a new method of best attributes reduction based on Rough Set theory.In this paper the method is utilized to solve the problem of fault diagnosis for power transformer for the first time. Linked with the method of Rough Set value reduction,a group of minimal decision rules are produced.Accordingly,the coding workload is greatly minished,attributes compounding queries and the problem of rules matching in the case of lacking key attributes are avoided,and so computing speed will pick up.Moreover, the effectiveness of this model can be enhanced by the self-study method,which is achieved by modifying the decision table through richening training sample. Finally,the case studies show that the method is simple and effective.