无迹粒子滤波跟踪的发动机性能退化预测  

Engine Performance Degradation Prediction Based on Unscented Particle Filter Tracking

作　　者：李耀华[1] 沈泽汉 LI Yao-hua;SHEN Ze-han(College of Aviation Engineering Civil Aviation University of China,Tianjin 300300,China)

机构地区：[1]中国民航大学航空工程学院,天津300300

出　　处：《机械设计与制造》2024年第9期378-384,共7页Machinery Design & Manufacture

基　　金：国家自然科学基金委员会--中国民航局民航联合研究基金重点支持项目(U2033209)。

摘　　要：针对航空发动机性能退化预测问题,这里提出了一个结合多阶段有效传感器融合的无迹粒子滤波发动机剩余寿命预测模型。模型针对数据融合,基于累积方差贡献率动态更新各个退化阶段的有效传感器,得到更符合实际情况的退化状态健康指标,然后应用无迹粒子滤波算法更新模型时变参数以实现发动机健康指标和剩余寿命的提前预测。这里采用发动机性能退化仿真实验数据集进行数据分析和验证。结果表明,所提出的基于方差贡献率的分阶段非线性融合方法,可以减少退化量中的冗余数据影响。所引用的无迹粒子滤波算法,有效解决了传统滤波算法中的粒子退化、结果发散问题。相较于一般的滤波方法在最终预测结果上,绝对误差减少7个周期,精度提高了3.6%,加强了粒子的跟踪能力,提高了目标跟踪的精度。为航空发动机的健康管理和可靠性评估提供参考依据。Aiming at the performance degradation prediction problem of aero-engine,a model for predicting the residual life of an engine based on untraceable particle filter combined with multi-stage effective sensor fusion is proposed.In view of data fusion,effective sensors at each degradation stage are dynamically updated based on cumulative variance contribution rate to obtain health indicators of degradation state that are more consistent with the actual situation.Then,the time-varying parameters of the model are updated by the untraceable particle filter algorithm to realize the advance prediction of engine health indicators and remaining life.the data set of engine performance degradation simulation experiment is used for data analysis and verification.The results show that the proposed multi-stage nonlinear fusion method based on variance contribution rate can reduce the influ⁃ence of redundant data in the degradation.The proposed algorithm effectively solves the problems of particle degradation and re⁃sult divergence in traditional filter algorithms.Compared with the general filtering method,the absolute error in the final predic⁃tion results is reduced by 7 cycles,and the accuracy is improved by 3.6%,which strengthens the tracking ability of particles and improves the accuracy of target tracking.It provides a reference for the health management and reliability evaluation of aeroen⁃gine.

关 键 词：方差贡献率 非线性融合 性能退化预测 无迹粒子滤波 航空发动机 

分 类 号：TH16[机械工程—机械制造及自动化] V239[航空宇航科学与技术—航空宇航推进理论与工程]