摘要
结合工程设备系统故障发生具有随机性的特点 ,基于概率粗集 (PRS)理论模型实现了故障诊断知识的提取。通过改进的模糊 C-均值聚类算法对原始故障数据加以量化 ,采用正则条件熵进行诊断知识系统的统计约简 ,对不协调的诊断规则利用最小风险 Bayes决策理论加以分析处理。仿真实验表明 ,该方法弥补了现有方法的不足 ,克服了基本 RS模型方法无法处理不协调规则的缺点 。
A PRS-model-based method of fault diagnostic knowledge extraction is proposed, allowing for the stochastic property of faults. Regular conditional entropy is used to make the statistic reduction, and the modified fuzzy C-mean clustering algorithm to discrete the raw data. Bayes decision under minimum risk is utilized to deal with the contradictory diagnostic rules. Simulation results for rolling bearings show that the method compensates the shortcomings of other existing approaches and the defects of the RS-based-method, and produces the effective simplified diagnostic knowledge.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2004年第5期600-603,共4页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金重点项目 ( 60 2 3 40 10 )资助
