Remaining Useful Life Prediction Of Turbofan Engine Based On Data-driven

Turbofan engine is a kind of thermal machinery with high precision.This kind of mechanical system has complex structure and often works in an extreme environment,which is prone to get out of order and cause an irreparable loss.In order to ensure the safe and reliable operation of the system and reduce the occurrence of disaster accidents,it is of great important significance to predict its remaining useful life.Due to the complexity of the engine working system,the physical model-based method which relies too much on prior knowledge is too difficult to implement and has poor applicability.Aiming at these problems,a data-driven method is used to model the degradation process of the mechanical equipment by using the data monitored by sensors installed on the mechanical equipment,so as to realize the prediction of the remaining service life of the engine.The specific contents of the study are as follows:The improved gradient boosting decision tree(GBDT)algorithm is introduced into the methodology for remaining useful life prediction.The improved GBDT is suitable for building models with higher featured dimensions.The turbofan engine performance data recorded by modern machines are normalized and used for improved GBDT training and testing,so as to achieve the purpose of reasonably predicting the remaining useful life of aero-engine.Compared with the traditional method,such as support vector regression(SVR),convolutional neural network(CNN),and combine of convolutional neural network and long short-term memory(LSTM-CNN),the performance has been significantly improved.A life prediction platform based on MFC is designed and developed.The data collected by the sensor is uploaded to the server.After the background technician obtains the data from the server,based on the research model of this article,the remaining useful life prediction is realized.The above research results show that the improved GBDT model proposed in this paper is applicable to the prediction of the remaining useful life of turbofan engine under different operating conditions.The proposed model reduces the dependence of traditional diagnostic methods on domain expert knowledge to a certain extent.Comparative experiments on the C-MAPSS data set have achieved remarkable and delightful results.