Research On Maintenance Decision-making Of Turbofan Engines Based On Attention-based Temporal Convolutional Network And Evolutionary Game

Turbofan engines are one of the most important parts of aircrafts.In order to ensure normal operation of the engines,it is necessary to carry out well maintenance management of turbofan engines.Reasonable maintenance decisions can not only improve service levels of maintenance service suppliers,but also improve economic benefits for maintenance service demanders.Predicting remaining useful life provides an important reference and basis for maintenance decision-making.Effective remaining useful life prediction can provide more accurate information for decision-making.Remaining useful life(RUL)prediction and maintenance decision-making are studied.The main research contents are as follows:(1)RUL prediction of turbofan engines is studied.An Attention-based Temporal Convolutional Network(ATCN)is proposed for RUL prediction.In order to enable the prediction model to capture the key information in features efficiently,two attention mechanisms,a self-attention mechanism and a squeeze-and-excitation mechanism,are introduced into a Temporal Convolutional Network to improve feature capturing abilities and RUL prediction performances of the model.In order to verify effects of the proposed ATCN,experiments are carried out on the C-MAPSS turbofan engine dataset.Results of the proposed ATCN are compared with those of other methods.Results show that the proposed ATCN is superior to other methods in terms of the root mean square error and Score.Effects of the self-attention mechanism and the squeezeand-excitation mechanism in ATCN are verified by analyzing samples.(2)In order to make a reasonable maintenance decision,balance interests of the maintenance service suppliers and maintenance service demanders,an evolutionary game model,in which payoffs of the maintenance service suppliers and maintenance service demanders are related to maintenance service level and the total value of turbofan engines in a decision-making cycle respectively,is established.The maintenance service suppliers make decisions on maintenance thresholds.The maintenance service demanders make decisions on whether turbofan engines continue to be maintained according to maintenance value and depreciation value of turbofan engines.In order to verify the established game model,a case study is designed.Pure strategy Nash equilibrium of the game model is solved and analyzed.Then,evolutionarily stable strategies of the game model are solved and analyzed through simulation.Finally,effects of two important parameters in the established evolutionary game model,the price of the turbofan engine and the maximum RUL of the turbofan engine at the initial time of a decision-making cycle,on the evolution of evolutionarily stable strategies are analyzed.Relevant maintenance decision-making suggestions are obtained.Results show that maintenance service suppliers should adopt a higher maintenance threshold as a strategy for the turbofan engine with a high price or large maximum RUL,and vice versa.