The change trend of exhaust gas temperature margin (EGTM) of civil aviation engine reflects its performance degradation. In order to employ the change trend of EGTM to map the performance degradation of the engine, an deep integrated neural network prediction method based on empirical mode decomposition (EMD) algorithm and convolution short-term memory network (CNN-LSTM) is proposed in this study. EMD is used to decompose the original EGTM sequence into multiple natural mode components and residual components, and all the resulting components are used as model input. The convolution long-term and short-term memory network is used to capture the nonlinear correlation of each component and extract the long-term dependence to construct the framework of deep learning model. In order to verify the effectiveness of the proposed method, the actual test data of EGTM of an airline for 10 years are used for experimental analysis, and five kinds of neural networks are designed as competitive models for comparative study. The experimental results show that the proposed EMD-CNN-LSTM fusion neural network model can reduce the mean absolute error and the root mean square error by 37.82% and 33.01%, and increase the goodness for fit by 1.02%, compared with competitive models. Furthermore, when EGTM is in the sensitive area, EGTM single-point prediction accuracy of the proposed model is significantly higher than that of other competitive models. Therefore, the proposed fusion neural network model has good accuracy and stability in EGTM prediction of civil aviation engine.