摘要
针对配电网发生故障后故障诊断警报信息存在不确定性和不完整性导致难以得出准确诊断结果的问题,提出一种基于粗糙集与决策树的配电网故障诊断算法,实现了对故障样本决策表进行无教师的规则提取。该算法将配电网的原始样本集转化成决策表,利用粗糙集具有较强的处理不确定和不完备信息的能力,对决策表的条件属性进行约简处理;同时,利用决策树具有快速学习及分类的优势对约简后的决策表进行诊断规则提取;将产生的规则运用于配电网故障诊断中以实现快速故障诊断。该算法提高了配电网故障诊断的精度和鲁棒性,最后通过算例验证了该算法的有效性。
To improve the indeterminacy and imperfection of distribution networks fault information that make it difficult to obtain accuracy fault diagnosis results, a new fault diagnosis algorithm based on the rough sets and decision tree theory is proposed, which can extract diagnosis rules directly from reduced decision table. The rough sets theory as a new mathematical tool is used to deal with inexact and uncertain knowledge for pattern recognition. The target is mainly to remove redundant information and to seek for reduced decision tables by use of discernible matrix. As a quickly learning theory and classification tool, according to the value of the information entropy, the decision tree is used to extract diagnosis rules directly from reduced decision table. The rules from decision tree can be with clear relationship and easy to explain. With the proposed method the blindness and redundancy of ruling can be avoided and the space for decision table is evidently diminished. According to the value of information entropy, the priority of different characteristic information can be conducive to acquiring quick and satisfactory results for distribution networks without looking up the whole rules. In addition when the fault information is imperfect, the results can be still acquired according to the remained information, so the proposed method possesses strong tolerant ability. This method is thus developed to ensure diagnosis precision and speed up the implementation of distribution fault diagnosis system. Finally, examples are given to verify its effectiveness, and the comparison with the former method shows its advantages.
出处
《高电压技术》
EI
CAS
CSCD
北大核心
2008年第4期794-798,共5页
High Voltage Engineering
关键词
配电网
故障诊断
粗糙集
约简
决策树
算法
distribution networks
fault diagnosis
rough sets
reduction
decision tree
algorithm
