Research On Prediction Of Remaining Useful Life For The Bearing Of Air Turbine Starter Based On Particle Filter

Air turbine starter(ATS)is the core device of the civil aircraft engine’s starting system.The bearing has a decisive effect on the reliability of the ATS’s operation as the main power transmission component of the ATS,which remaining useful life(RUL)prediction has always been the focus of attention.Particle filter(PF)has been used in a variety of engineering field for the problem of remaining useful life prediction widely as a method of state estimation for the nonlinear system,showing superior performance.Using the particle filter to realize the prediction of the remaining useful life of the bearing accurately has great engineering value and theoretical significance.This paper has mainly completed the following four aspects of research:Firstly,establish a physical model of the degraded state of the bearing.In the terms of the state tracking and modeling of ATS bearings,the implicit feature is extracted from the vibration acceleration signal of the bearing,and the concept of degradation factor is put forward to map the degradation state of the bearing.The remaining useful life of ATS’s bearings is defined according to the degradation factor.Then,a physical model of the state degradation of the bearing is established according to the classic crack fatigue growth model.At the same time,the least square method is used to update the parameters in the physical model to make it more suitable for the actual degraded state of the bearing in real time.Secondly,propose a new particle filter algorithm.Based on the physical model of the degradation state of the bearing’s performance,aiming at the particles’ degradation and depletion problems that are common in the particle filter algorithm,learning from the idea of swarm intelligence optimization,the fireworks algorithm,which has been improved from the aspect of explosion strategy,mutation operator,mutation and selection strategy is introduced to redesigned the resampling process of particle filter.On this basis,a new particle filter algorithm,Fireworks Algorithm-Based Particle Filter(FPF)is formed.FPF algorithm can suppress the problem of particles’ degradation and depletion effectively.Then,develop the information management system of the bearing.The environment of fatigue acceleration experiment for the bearing of ATS is designed,the framework of hardware and software of information collection for the bearing is built,and a multi-dimensional information management system of the bearing is developed,which can realize the real-time collection and post-processing of bearing state information.On this basis,the fatigue acceleration experiments of four groups of bearings are completed.Finally,carry out experimental evaluation.Based on the bearing degradation model,improved particle filter and the collected valid data,the algorithm framework of remaining useful life prediction for the bearing of the air turbine starter based on particle filter is verified.Implement the remaining useful life prediction of four sets of bearings effectively,obtain specific prediction results and compare them with the prediction results of existing algorithms.The experimental results show that the average error rate of the prediction framework proposed by this subject is below 8.4%,indicating that the method has better performance.