基于MoG-HMM的齿轮箱状态识别与剩余使用寿命预测研究

提出了基于混合高斯隐马尔可夫模型的齿轮箱状态识别与剩余使用寿命预测新方法。建立了基于聚类评价指标的状态数优化方法，通过计算待识别特征向量的概率值来识别齿轮箱当前状态。在状态识别的基础上，提出了剩余使用寿命计算方法。最后，利用齿轮箱全寿命实验数据进行验证，结果表明，该方法可以有效的识别齿轮箱状态并实现了剩余使用寿命预测，平均预测正确率为90.94%，为齿轮箱的健康管理提供了科学依据。

Gearbox state identification and remaining useful life prediction based on MoG-HMM

A new approach for state recognition and remaining useful life (RUL) prediction based on Mixture of Gaussians Hidden Markov Model (MoG-HMM) was presented. State number optimization method was established based on cluster validity measures. One can recognize the state through identifying the MoG-HMM that best fits the observations. Then, the RUL prediction method was presented at the recognition base. Finally, the data of gearbox’s full life cycle test was used to demonstrate the proposed methods. The results showed that the mean accuracy performance was 90.94%.