摘要
针对传统轴承性能退化评估方法中退化阶段划分的主观性以及连续隐马尔科夫模型在建立评估模型时只考虑正常状态下的样本所引起评估结果的不足,提出了一种基于连续隐马尔科夫模型的轴承性能退化程度综合评估方法。该方法首先通过支持向量聚类方法将轴承全寿命周期划分成若干个退化阶段,然后从每个阶段中提取一定比例的样本用于训练,采用轴承正常阶段的训练样本建立轴承的连续隐马尔科夫模型,将不同退化阶段的训练样本输入模型,分别得到不同阶段样本相对于所建立正常阶段的连续隐马尔科夫模型的输出概率,据此得到样本隶属于不同退化阶段的隶属函数分布。最后,采用集对分析的方法建立轴承测试样本相对于正常阶段样本的联系度,并最终得到轴承性能退化程度的综合得分。通过利用轴承全寿命数据,并与传统连续隐马尔科夫模型及传统无量纲指标进行了对比,验证了所提出的综合评估方法在轴承性能退化评估方面的有效性。
Aiming at the deficiencies that the degradation stage division in the traditional evaluation method for bearing performance degradation has subjectivity, and in establishing the evaluation model the CHMM (continuous hidden Markov model) only considers the evaluation result caused by the samples under normal stale, this paper proposes a comprehensive evMuation method for bearing performance degradation based on CHMM (continuous hidden Markov model). In this method, firstly, the support vector clustering method is used to divide the full life of the bearing into several degradation stages, then a certain proportion of the samples fi'om every divided degradation stage are extracted and used as the training samples. Moreover, the training samples of normal stage are adopted to establish the CHMM of the bearing. The training samples in different degradation stages are input to the model. The output probabilities of the samples from different degradation stages relative to the established CHMM under normalstage are obtained, respectively, based on which the membership function distribution of the samples belonging to different degradation stages is obtained. Finally, the set pair analysis method is adopted to establish the correlation of the beating test samples relative to the sample in normal stage, and eventually the comprehensive score of the bearing performance degradation is obtained. The proposed method is tested with the complete life data of the bearing, and compared with traditional CHMM and traditional dimensionless indices, and the results verifythe effectiveness of the proposed comprehensive evaluation method in bearing performance degradation evaluation.
作者
姜万录
杨凯
董克岩
张生
Jiang Wanlu Yang Kai Dong Keyan Zhang Sheng(Hebei Provincial Key Laboratory of Heavy Machinery Fluid Power Transmission and Control, Yanshan University, Qinhuangdao 066004, China Key Laboratory of Advanced Forging & Stamping Technology and Science, Ministry of Education of China, Yanshan University, Qinhuangdao 066004, China)
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2016年第9期2014-2021,共8页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(51475405)
国家重点基础研究发展计划(973计划)(2014CB046405)项目资助
关键词
连续隐马尔科夫模型
综合评价方法
集对分析
支持向量聚类
continuous hidden Markov model (CHMM)
comprehensive evaluation method
set pair analysis
support vector clustering