11 / 2021-06-08 18:59:15

Multi-task Learning Based Classified-assisted Prediction Network for Useful Life Prediction

The turbofan engine is an important part of the aircraft. In order to provide proper maintenance to the turbofan engine to improve the reliability of the system, it is necessary to estimate the remaining useful life (RUL) of the engine. This paper proposes a classification-assisted prediction network (CAPN) through the multi-task learning method, which effectively realizes RUL prediction. First, in order to better complete the fault recognition task and extract fault feature information, Atrous Convolutional Network (ACNN) is used as the auxiliary task network. Secondly, ACNN-Long Short Term Memory(LSTM) is used as the main task network to realize RUL prediction. Finally, The interaction between multiple tasks is realized by designing a special Loss Function. In order to prove the effectiveness of this method, experiments were conducted on the C-MAPSS dataset provided by NASA, and high prognostic accuracy was obtained. The comparison with the existing methods using the same data set illustrates the broad prospects of this method in industrial applications.