基于重要度的约简算法在电网故障诊断中的应用

REDUCTION OF ATTTIBUTE IMPORTANCE ALGORITHM APPLICATION IN POWER DISTRIBUTION NETWORK FAULT DIAGNOSIS

为了解决不一致决策表的问题,在原有经典依赖度算法基础上,提出了一种基于重要度的约简算法。约简后的决策表保留了与原始决策表一样的决策属性,但其条件属性较原始决策表缩减了很多。通过具体实例,对大规模电网中的原始不一致决策表进行重要度的计算,根据重要度进行约简,得到去除冗余属性后的决策表,较好地解决了不一致决策表因信息缺失造成决策属性失真的问题。最后指出了粗糙集理论研究中现存的不足,并对如何克服不足提出了具体的思路。

Witli the purpose of solving the inconsistence of decision tables,basing on the classical dependability algorithm,the attrilD-ute importancc-based decision table reduction algorithm are proposeed. The reduccd decision table retains the same decision attributes with the original decision table,but the condition attributes are reduccd a lot . In this paper, specific examples are given to illustrate how to calculate the importance degree of the original inconsistent decision table of large-scale power grid. In accordance withi the im-portance degree, redundant attributes is removed to get a decision table, which help to solve the problem of tributes caused by the missing information in inconsistent decision tables. The existing problems of out, and some concrete thoughts on how to overcome the shortcomings are puted forward.