基于无限隐Markov模型的旋转机械故障诊断方法研究

针对传统隐Markov模型(HMM)在机械故障诊断中存在的不足,即HMM过学习或溢出问题以及隐状态数需要事先假定,提出了基于无限隐马尔可夫模型(i HMM)的机械故障诊断方法。在提出的方法中,以谱峭度为特征提取,i HMM为识别器,并以最大似然估计来确定设备运转中出现的故障类型。同时,将提出的方法与传统的HMM故障识别方法进行了对比分析。实验结果表明,提出的方法是有效的,得到了非常满意的识别效果。提出的方法能够有效避免了HMM在建模初期遗留下的不足,可以自适应确定模型中隐藏状态数和模型数学结构,因此,提出的方法明显优于HMM故障识别方法。

Research on rotating machinery fault diagnosis method based on infinite hidden Markov model

Aimingat the deficiency of traditional HMM fault recognition model in machinery fault diagnosis, i.e. over learning or overflow problems and requiring to assume the hidden states in advance, a new machinery fault diagnosismethod based on infinite Hidden Markov Model (iHMM) is proposed. In the proposed method, the spectral kurtosis is used as the fault feature extraction, the iHMM as theidentifier, and the maximum likelihood estimation is used to determinethe mechanical fault type occurred in the equipment operation. At the same time, the proposed method and traditional HMM fault identification method are compared and analyzed.The experiment result shows that the proposedrecognition method has very satisfactory recognition effect. The proposed method can effectively avoid the deficiency of the HMM method in the initial modeling stage, can adaptively determine the number of hidden states in the model and the mathematical structure of the model. Therefore,the proposed method is obviously superior to the traditional HMM fault recognition method.