Research On Aero Engine Maintenance Strategy Based On Reinforcement Learning

During the operation of the aircraft,engine equipment may undergo state decay due to fatigue,wear,aging,etc.Aero-engines with degraded operating conditions may lead to a decline in aircraft safety performance and increase aircraft maintenance costs.In order to prevent aircraft engines from operating in poor conditions,appropriate maintenance actions can be taken,such as inspection,repair or replacement,etc.However,if the aircraft engine is overmaintained,it may interfere with the operation and increase the downtime.Therefore,it is very important to develop a reasonable maintenance strategy for aircraft engines.This paper mainly uses reinforcement learning methods to study aircraft engine decay problems.According to the condition monitoring information,event information and maintenance information during the operation of the aircraft engine,the aircraft engine condition evaluation model is established to identify some abnormal parameters during the operation of the aircraft engine,and to provide support for maintenance decisions to extend the service life of the engine.Use the Markov decision model in the discrete state to characterize the state of the aircraft engine,and establish the aircraft engine recovery model under imperfect maintenance.Finally,for a research example,the Q-learning method is used to solve the decay model to obtain the average return of the aircraft engine and the predictive maintenance strategy based on the condition,which has certain practical guiding significance.