An Ensemble Learning Prognostic Method for Remaining Useful Life Prediction of Aircraft Engines
首发时间:2020-11-03
Abstract:Data-driven models have been widely used to predict the remaining useful life (RUL) of many engineering systems, e.g. aircraft engines. However, two shortcomings exist: (i) single algorithm has performance limitations for the specific application and (ii) reliably tracking the degraded performance of aircraft engines remains challenging. In this paper, a new ensemble learning prognostic method is proposed, which considers the effects of performance degradation on RUL. First, the overall degradation process is divided into multiple degradation stages, which present the performance of aircraft engines. Then, in each degradation stage, the higher prediction accuracy the base learner obtains, the higher weight is assigned to the base leaner. Finally, based on the obtained weights, the predicted results of all base learners are combined to predict the RUL of aircraft engines. The experimental results of aircraft engines verify the effectiveness and practical value of the proposed method. The results show that the proposed method has a good predictive effect and fits the degradation curve of the aircraft engines well.
keywords: General industrial technology Reliability Ensemble learning Remaining useful life Aircraft engine prognostics
点击查看论文中文信息
基于集成学习的航空发动机剩余使用寿命预测
摘要:针对航天发动机退化数据的时变复杂性特点,提出了一种基于集成学习的预测方法,同时考虑了性能退化对剩余使用寿命(RUL)预测的影响。首先,使用KNN聚类算法将整个退化过程分为多个退化阶段。然后,在每个退化阶段,分配给训练过程中预测精度高的基学习者较高的权重。最后,根据得到的权值,将所有基学习者的预测结果进行加权,从而对飞机发动机的RUL进行预测。使用航空发动机的试验数据验证了该方法的有效性和实用价值。结果表明,该方法具有良好的预测效果,能很好地拟合飞机发动机的退化曲线。
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
基于集成学习的航空发动机剩余使用寿命预测
评论
全部评论0/1000