Aero-engine Exhaust Gas Temperature Prediction Based on the Deep Convolution Neural Network and LightGBM Combination Model

Exhaust gas temperature (EGT) is an important characteristic parameter inside aero-engine. The prediction of EGT is of great significance for improving the performance and safety of aero-engine. This paper proposes a model combining a deep convolution neural network (DCNN) and a light gradient boosting machine (LightGBM) algorithm to estimate the EGT. Unlike traditional prediction and health management (PHM) technologies, it does not require professional signal processing knowledge and complex mathematical models. The design procedure is shown as follows. First, the time window of the original Quick Access Recorder (QAR) data is normalized and used as the input of the DCNN, and the DCNN is devoted to extract useful information from input data. Second, in order to avoid the limitations of the fully connected layer, we replaced the DCNN fully connected layer with a strong classifier LightGBM to improve the accuracy of the prediction. Finally, we conducted some experiments on QAR data, and the results verify that the proposed combined model of DCNN and LightGBM has a higher accuracy rate compared with other traditional algorithms.