EGTM Data-Driven Aero-Engine Washing Schedule Optimization Using Hybrid Prediction Model

Effective planning of aero-engine washing schedules can significantly reduce operating cost of airlines and greatly extend the engine's service life. The engine exhaust gas temperature margin (EGTM) is a crucial indicator for assessing the health of engine, and developing a reasonable maintenance program. However, the significant nonlinearity of EGTM affected by many factors presents challenges for long-term predictions. Specifically, the aero-engine washing will result in a stepwise jump in EGTM, thus, it is difficult to predict the EGTM evolution using traditional methods. This article proposes a hybrid prediction model that integrates the long short-term memory, the autoregressive integrated moving average, complete ensemble empirical modal decomposition with adaptive noise, and the Markov Chain (Markov) models. An EGTM linear decay model is first constructed to assess the recovery effect of EGTM after washing. Then, a transfer learning model is employed to predict model parameters. Finally, an engine cleaning schedule optimization method is proposed to realize EGTM prediction as well as the washing effect prediction. The effectiveness of the proposed approaches is demonstrated on the real data from the CFM56-7B engine.