基于粗糙决策树模型的复杂设备智能故障诊断

提出一种粗糙决策树智能故障诊断方法。首先利用粗糙集理论构建故障诊断决策表；然后应用最小熵离散算法对特征数据进行离散；接着利用基于分辨矩阵的算法对决策表进行属性约简，并基于最大聚类比原则优选出最佳约简；最后利用C4.5算法构造出诊断决策树。并对实例进行故障诊断。结果表明：该方法不仅能有效地减少特征数据获取的工作量，而且能快速准确地实现故障识别，具有比粗糙集诊断法、粗糙决策诊断法和C4.5诊断法更强的工程实用性。

Intelligent Fault Diagnosis Method for Complex Equipment Based on Rough Decision Tree Model

A rough decision tree-based intelligent fault diagnosis method for complex equipment was proposed. The original diagnostic decision table was built by using the rough set theory; the continu?ous conditional attributes were discretized by using the minimum entropy-based algorithm; the redun?dancy conditional attributes were reduced by utilizing the discernibility matrix-based algorithm, and the optimal attribute reduction is established by using the maximum cluster ratio principle; the diag?nostic decision tree was constructed by utilizing the C4.5 algorithm. The fault diagnosis of an example was made by the proposed method. The diagnosed results show that the proposed method can effec?tively decrease the workload of feature datum extraction, recognize the fault patterns rapidly and accu?rately, has better engineering practicality than the rough set-based, the rough decision-based and the C4 . 5-based metKod.