论文部分内容阅读
剩余寿命(RL)预测是设备预测维护的关键环节。准确在线预测能够为维护策略的实时安排提供更加精确的技术支持,有效避免失效的发生。工程实际中,反映设备退化过程的性能指标往往不能直接监测,为解决隐含退化过程的剩余寿命在线预测问题,提出一种基于半随机滤波-期望最大化(EM)算法的预测方法。首先以剩余寿命为隐含状态,构建状态空间模型描述直接监测数据与设备剩余寿命间的随机关系。为实现单个设备剩余寿命的在线预测,依据到当前时刻为止的监测数据,采用扩展卡尔曼滤波(EKF)与期望最大化算法相互协作的方法实时估计与更新模型未知参数和剩余寿命分布。最后,将该方法用于惯性测量组合(IMU)剩余寿命在线预测问题,实验结果表明该方法能够提高预测的准确性并减少预测的不确定性。
Remaining life (RL) prediction is a key part of predictive maintenance of equipment. Accurate online forecasting can provide more precise technical support for the real-time arrangement of maintenance strategies and effectively prevent the failures from occurring. In engineering practice, the performance indexes that reflect the process of equipment degeneration can not be directly monitored. To solve the problem of online service life prediction of hidden degenerative process, a semi-random filter-expectation maximization (EM) prediction method is proposed. Firstly, the residual life is implied and the state space model is constructed to describe the stochastic relationship between the direct monitoring data and the remaining life of the equipment. In order to realize the online prediction of the remaining life of a single device, based on the monitoring data up to the present moment, EKF and expectation maximization algorithms are used to estimate and update the unknown parameters and residual life distribution in real time. Finally, the method is applied to the residual life prediction of IMU. The experimental results show that this method can improve the accuracy of prediction and reduce the uncertainty of prediction.