Remaining Useful Life Prediction of Turbofan Engines Using CNN-GRU: A Comparative Study on C-MAPSS Dataset

Abstract

Accurate prediction of Remaining Useful Life (RUL) is crucial for effective predictive maintenance strategies in various industries. This study investigated the application of a combined Convolutional Neural Network (CNN) and Gated Recurrent Unit (GRU) architecture for predicting the RUL of turbofan engines using the C-MAPSS dataset. The CNN-GRU model leverages the complementary strengths of both components, with the CNN extracting spatial features from sensor data and the GRU capturing temporal dependencies in sequential engine data. Preprocessing techniques such as windowing and data reshaping were employed to optimize the input for the deep learning model. Six CNN-GRU configurations were tested to identify the optimal architecture for each C-MAPSS subset (FD001, FD002, FD003, and FD004). The model performance was evaluated using Root Mean Squared Error (RMSE) and Mean Squared Error (MSE) metrics. The results demonstrated that the CNN-GRU model effectively captured degradation patterns and accurately predicted RUL across all datasets, outperforming traditional machine learning methods, such as Multilayer Perceptron (MLP) and Support Vector Regression (SVR). The robustness and adaptability of the model to various operational scenarios highlight its potential for implementation in diverse industrial applications, enhancing predictive maintenance efforts, and improving overall operational efficiency. Future research should explore alternative configurations and consider increasing the number of training epochs to refine the prediction results further.