考虑三源不确定性的Wiener过程航空发动机剩余使用寿命预测

Prediction of remaining useful life time of aero engine based on the Wiener process of three-source uncertainty

针对航空发动机退化过程中存在三源不确定性,与现有研究假设测量误差为正态分布的问题,提出测量误差为Logistic分布的Wiener过程航空发动机剩余使用寿命(residual useful life,RUL)预测方法。首先,建立测量误差服从Logistics分布时的性能退化模型,并在首达时间下给出RUL分布的准确数学表达式。其次,为提高监测参数的利用率,给出一种基于Spearman系数的关键参数筛选方法。最后,针对缺少历史数据和先验信息的问题,采用基于最大熵无迹粒子滤波和ECM算法的参数估计方法,同时在得到新的性能退化数据之后,能对模型参数自适应更新。采用均方误差(mean squared error,MSE)作为评价指标。实验结果表明:该方法的MSE值为13.25,低于传统方法的MSE值16.12,可以有效地提高预测精度、发动机使用率和安全性。

Given the problem of three-sources uncertainty in aero engine degradation process and the assumption of normal distribution of measurement error in existing research,a Wiener process aero engine residual useful life time(RUL)prediction method with logistic distribution of measurement error was proposed.Firstly,the performance degradation model was established when the measurement error obeys the logistics distribution,and the accurate mathematical expression of the RUL distribution was given under the first arrival time.Secondly,in order to improve the utilization of monitoring parameters,a key parameter screening method based on Spearman coefficient was given.Lastly,in view of the lack of historical data and prior information,a parameter estimation method based on maximum entropy unscented particle filtering and conditional expectation maximization algorithm was adopted.Meanwhile after new performance degradation data was obtained,the model parameters could be adaptively updated.With MSE as the evaluation index,the experimental results show that the MSE value of this method is 13.25,and the MSE value is 16.12 lower than that of traditional method,which can effectively improve the prediction accuracy and the engine utilization and safety.