Combining Autoencoder with Similarity Measurement for Aircraft Engine Remaining Useful Life Estimation

Abstract

Remaining useful life (RUL) estimation is very important for the maintenance of aircraft engines. We can evaluate the current condition of an aircraft engine and predict its RUL by constructing its degradation curve. However, the degradation curve is often difficult to obtain due to the unobserved degradation patterns. Currently, many researchers estimated RUL by setting a fixed RUL target function or making assumptions about how an engine degrades. But in the real world, degradation models of aircraft engines are generally individualized. To obtain personalized degradation curves and predict RUL accurately, we propose a method based on autoencoder and similarity measurement. First, an autoencoder trained with normal data is adopted to extract degradation curves of aircraft engines and build a degradation model template library. Then, we measure each test object with all template curves to get similarities and corresponding RULs based on a sliding window and complexity-invariant distance. At last, the estimated RUL can be obtained by calculating the weighted average of highly relevant corresponding RULs. We conduct the proposed method on the aircraft engine dataset provided by NASA. The experimental results demonstrate that our method can utilize the information of multi-sensor data to generate personalized degradation curves effectively and estimate RUL more accurately.