Health Status Assessment of Aero-Engine Based on Multilayer Information Sharing and MOSE

As the heart of an aircraft, the complex structure of an aero-engine poses a considerable challenge for its maintenance. The traditional health status assessment includes two tasks of remaining useful life prediction and health status division, but this paper adds the study of reliability assessment of aero-engine operation process on its basis and proposes an aero-engine health status assessment method based on multilayer information sharing and MOSE. Firstly, a series of machine learning methods such as principal component analysis are used to preprocess the data, and a label construction method based on slope division is adopted. Secondly, the shared information in the sensor signal is extracted using the MOSE network module and the extracts task-specific information in the sensor signal is extracted by the BIGRU network module, and the weighted decision is realized by the gating network on the output of both, and the prediction output of each sub-task is performed by the sub-task module. Finally, experiments on the C-MPASS dataset show that the proposed method in this paper outperforms recent popular models.