基于最优阶次HMM的电机故障诊断方法研究  

Research on motor fault detection method based on optimal order hidden Markov model

作　　者：胡为[1] 高雷[2] 傅莉[1] 

机构地区：[1]沈阳航空航天大学航空航天工程学部,沈阳110136 [2]中国科学院沈阳自动化研究所信息服务与智能控制技术研究室,沈阳110016

出　　处：《仪器仪表学报》2013年第3期524-530,共7页Chinese Journal of Scientific Instrument

基　　金：国家863计划专题课题(2008AA042901);国家自然科学基金(61074090);中航创新基金(cxy2011SH)资助项目

摘　　要：介绍了一种基于隐马尔可夫模型的电机匝间短路故障诊断方法,方法中提出了一种基于信息熵的隐马尔可夫模型阶次选取策略,该策略根据求解符号出现概率的熵速率与隐马尔可夫模型阶次的关系确定模型最佳阶次,确保在得到合理的模型结构的前提下,使该模型的计算代价最小,从而得到最优隐马尔可夫模型。将该最优隐马尔可夫模型应用于电机匝间短路故障诊断实验中,获得满意的诊断结果,结果表明在保证匝间短路故障诊断精度不变的情况下,通过合理选取隐马尔可夫模型的阶次可以有效地减少模型的计算代价,提高模型的计算效率。A fault detection method of motor inter-turn short circuit based on Hidden Markov Model （HMM） is introduced in this paper,and an order selection strategy for Hidden Markov Model based on information entropy is proposed in the method. The optimal order of Hidden Markov Model is decided according to the relationship between the entropy rates of symbol occurrence probabilities and the order of Hidden Markov Model. A more accurate and more reasonable Hidden Markov Model is obtained with the selected optimal order;at the same time,the model calculation cost is minimized. The method proposed in this paper was applied to the fault detection experiment of motor inter-turn short circuit;and the experiment results indicate that under the condition of ensuring the fault detection accuracy of motor inter-turn short circuit, the model calculation cost can be decreased effectively by reasonably selecting the order of Hidden Markov Model.

关 键 词：隐马尔可夫模型 阶次选取 匝间短路 故障诊断 信息熵 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]