Prediction of the temperature of diesel engine oil in railroad locomotives using compressed information-based data fusion method with attention-enhanced CNN-LSTM

Engine oil temperature is a key parameter for ensuring the optimal functioning of diesel engines in locomotives. This paper proposes attention-enhanced CNN-LSTM based on compressed information for predicting engine oil temperature of railroad locomotives using sensor data associated with operational, system-related, and environmental factors. First, the autoencoder was utilized to transform the raw input into latent representation to reduce the computation cost and extract meaningful data patterns. Subsequently, the CNN (convolutional neural network)-LSTM (long short-term memory neural network) was adopted to capture short-term and long-term dependencies within the compressed data by combining the advantages of both CNN and LSTM architectures. CNN is to consider the local correlations, while LSTM is to learn the long-term sequential dependence in the data. Furthermore, the self-attention mechanism was incorporated to enhance the performance of CNN-LSTM by weighing the importance of different parts. An experiment has been conducted using real-world dataset. The results show that our proposed model can make highly accurate predictions with Mean Square Error (MSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) values of 0.335, 0.369, and 0.586%, respectively, which outperform five baselines (i.e., random forests, XGBoost, CNN, LSTM, and CNN-LSTM). This emphasizes the significance of considering both local and long-term correlation and integrating attention mechanism in handling sequence data, contributing to better prediction performance. Overall, the proposed approach provides accurate predictions of engine oil temperature and can be used to support the efficiency and reliability of in-service diesel engines in locomotives by optimizing operating conditions and allowing for proactive measures to prevent overheating before critical temperatures are reached.