Aeroengine After-Maintenance Performance Prediction Based on Simplified Mixed Takagi—Sugeno Model

Abstract

An aeroengine is a type of complex equipment, for which reasonable maintenance is essential to ensure the reliability of its life-cycle operation. The after-maintenance performance prediction method can provide infrastructure supports for aeroengine maintenance optimization. The aeroengine after-maintenance performance is affected by the before-maintenance performance and maintenance levels. The aeroengine performance is represented by a time series; however, the maintenance-level data are regarded as discrete variables. For these characteristics, a simplified structure identification and mixed variables Takagi–Sugeno (SMTS) model, which can simultaneously address time series and discrete variables, is proposed. The proposed model can predict the after-maintenance performance time series for the condition of a small sample set. The effectiveness of the proposed SMTS model is demonstrated by the real-valued data of an aeroengine fleet. The traditional neural network, process neural network, support vector regression, and deep learning network are adopted as comparison models. The results validate that the proposed SMTS model has the distinct advantages of prediction accuracy and stability.