A LS-SVM based Approach for Turbine Engines Prognostics Using Sensor Data

Abstract

As the core part of aircraft, the engine’s health condition is tightly related to the safety of the aircraft. Therefore, to predict the remaining useful life (RUL) of the engine accurately has a great practical significance. Since the turbine engine is a complex system and its life degradation process is affected by many factors, it’s hard to use a certain model to represent the engine degradation. Fortunately, the turbine engine often sets a variety of sensors to obtain operational data, which provides an effective way to monitor the trend of degradation. In this paper, a prognostic approach based on least squares support vector machine (LS-SVM) is proposed to estimate the turbine engine RUL using sensor data. Firstly, the health indicator (HI), which is a synthesis indicator from multivariate sensor data, is calculated using descending dimension method, then the degradation models are built through regressing the HIs based on LS-SVM. Finally, the RUL of the testing engines can be calculated by the degradation models. As an application case, the turbofan engine degradation simulation data set supplied by NASA Ames is used to demonstrate the performance of the proposed approach.