Transfer Learning‐Based Domain‐Adaptive One‐Dimensional Convolutional Neural Network for Fault Diagnosis of Rotating Machines

ABSTRACT In recent years, deep learning models, particularly one‐dimensional convolutional networks (1‐D CNNs), have shown significant potential for fault diagnosis of rotating machines. However, existing methods often struggle to generalize to real‐time data and lack adaptability across different operating conditions. To address these challenges, this paper proposes a transfer learning‐based domain‐adaptive 1‐D CNN framework. In this framework, the 1‐D CNN model is initially pre‐trained on source domain data and then fine‐tuned on the target domain by freezing the first three convolutional layers while updating the remaining layers to adapt to domain‐specific features. The proposed framework was validated using rolling bearing and real‐time wind turbine gearbox vibration data. The experimental results show a diagnostic accuracy of 99.99% on bearing fault datasets under varying load conditions, outperforming other state‐of‐the‐art transfer learning methods. Additionally, the model pre‐trained on bearing data achieved a diagnostic accuracy of 98.52% when applied to real‐time gearbox vibration data. These findings confirm the effectiveness of the proposed framework across different settings and its potential applications for a wide range of rotating machinery in the industry.