Vibration-based structural health monitoring via deep learning techniques: A review

The vibration signal is closely related to abnormal states, which is an effective diagnostic tool in structural health monitoring (SHM) fields. Deep learning methods have got remarkable success in the detection of damage through vibration signals. This paper presents a systematic review of deep learning methods for SHM that utilizes vibration signal data from different model perspective. In recent years, there has been a tremendous increase in the number of research involving deep learning for vibration-based structural health monitoring. The accuracy of such works is equivalent to that of traditional machine learning approaches, and a better result could be achieved by integrating multiple approaches. Furthermore, we found that transfer learning methods yield the promising results when limited data is utilized to train the model. This paper provides a comprehensive review of existing deep learning research in SHM using vibration signal data from multiple perspectives and fulfills the gap that existing reviews lack in the discussion of transfer learning for SHM by presenting the highlights of the transfer learning. Finally, we summarize the contents of the paper and analyze the challenges of existing research to make recommendations for future work.