Intelligent Fault Diagnosis of Wind Turbine Generator Bearings Using Acoustic Signals

Wind turbine generator bearings are key components in wind power generators, where accurate and convenient fault diagnosis for product predelivery inspection and depot repair inspection is a major challenge to ensure their safe operation. In this study, an artificial-intelligence-based method was developed for bearings fault diagnosis using acoustic signals with convenient capture, collection, and transmission. Specifically, the running sound was used as the input signal; then, several machine learning models and deep learning models were used to analyze the acoustic signals. The performance of deep learning models using vibrational signals collected by acceleration sensors was investigated for comparison. Results show that an accuracy of 99.90% can be achieved by utilizing deep learning models. The developed method will be a powerful tool for accurate and convenient fault diagnosis because it can be easily deployed on embedded devices. Empirical results validate the effectiveness of our proposed method.