Wind Turbine Gearbox Temperature Prediction Based on the DSCNN-BiLSTM Model

The health of the gearbox is critical for the safe and stable operation of wind turbines, with its operating temperature serving as a reliable indicator of gearbox condition. However, due to the non-linearity and complexity of temperature signals collected in real-world applications, traditional prediction methods often struggle to accurately forecast the actual operating temperature of wind turbine gearboxes. To address this challenge, this paper presents a novel deep learning model designed to predict temperature fluctuations in wind turbine gearboxes and evaluate their health status. The proposed model integrates a Deep Separable Convolutional Neural Network (DSCNN) and a Bidirectional Long Short-Term Memory (BiLSTM) network, combining strong spatial and temporal feature extraction capabilities. Experimental results demonstrate that the proposed method outperforms four existing models in terms of temperature prediction accuracy, exhibiting superior adaptability and sensitivity, thereby providing a more effective approach for gearbox health monitoring in wind turbines.