基于退化速率跟踪粒子滤波的剩余使用寿命预测框架

毋庸置疑,剩余使用寿命预测对于设备的健康管理越来越重要。近年来粒子滤波方法被越来越多地应用到设备寿命预测技术当中,这是因为粒子滤波方法能更好的解决非线性非高斯系统滤波问题,而且能够获得不确定度信息。但该方法的预测性能却过度依赖于预测模型,并且对于模型参数的初始分布也比较敏感,这在一定程度上限制了粒子滤波预测方法的进一步发展。本文针对基本粒子滤波预测方法的不足,提出了一种基于退化速率跟踪粒子滤波的通用预测框架,以历史观测数据的退化速率统计规律作为指导来跟踪目标数据的退化速率,实现对粒子滤波预测方法的简化。并将该方法用于轴承和锂离子电池的剩余使用寿命预测,验证了方法的有效性。

A framework for remaining useful life prediction based on degradation rate tracking particle filter

There is no doubt that remaining useful life prediction is important to the health management of modern equipment. Particle filter method has been widely applied to the prediction of equipment remaining useful life in recent years, because it can solve the filtering problem of nonlinear and non-Gaussian systems better and allow the uncertainty management. However, the prediction performance of a particle filter is largely dependent on the prediction model and very sensitive to the initial distribution of the model parameters. These flaws limit the further development of particle filter methods in the prediction to a certain extent. Aiming at the shortcomings of the basic particle filter prediction method, this paper presented a kind of general prediction framework of particle filter based on tracking degradation rate for predicting degradation trend. In the proposed method, the statistical rule of historical data was utilized to guide the tracking of degradation rate and simplify the prediction process. The remaining useful life prognosis case of rolling bearings and Li-ion battery verified the effectiveness of the proposed method.