Remaining Useful Life Prediction for Airborne Fuel Pump Based on AR and Particle Filter

In order to predict the remaining life of high reliability and long life airborne fuel pump, a fusion approach based on adaptive regression (AR) and particle filter (PF) is put forward. Firstly, the basic principles of adaptive regression algorithm and particle filter algorithm are introduced. Then, a fusion framework based on statistical prediction and data driven prediction is proposed. Secondly, taking into account the relationship between outlet pressure and performance degradation of airborne fuel pump, an AR model is used to establish the life degradation model of the airborne fuel pump. Finally, the RUL prediction of airborne fuel pump is realized by the fusion particle filter algorithm. The experimental results illustrate that compared to traditional regression approach, the proposed approach not only can effectively predict the life of airborne fuel pump, but also has the uncertainty expression of the prediction result.