基于随机滤波模型的齿轮箱剩余寿命预测研究

Study of Gearbox Residual Life Prediction based on Stochastic Filtering Modeling

针对设备剩余寿命预测无法获取设备直接状态信息的问题,引入随机滤波模型,利用平时易于监测到的间接状态信息,来预测设备的剩余寿命。该模型采用贝叶斯递推理论,可以有效利用设备监测到的历史状态信息;针对小样本模型参数估计问题,采用主观数据和客观数据相结合的贝叶斯方法对模型的参数进行估计;最后,以齿轮箱全寿命实验为依据,利用该模型对其剩余寿命进行预测,为复杂设备剩余寿命预测提供了新的研究思路。

To solve the problem of lacking direct condition information in predicting equipment residual useful life（RUL）,stochastic filtering model（SFM） is built for equipment＇s RUL prediction with indirect information,which is easy to get.SFM adopted Bayesian recursive theory which could fully use equipment＇s history condition information.For the estimation of model＇s parameters with insufficient samples,the Bayesian method is to update the parameters estimated by subjective data and objective information.Finally,the RUL prediction of SFM model＇s calculation process is carried out based on full life test for gearbox.A new way for state recognition of complex equipment is provided.