A survey of mechanical fault diagnosis based on audio signal analysis

Abstract

Mechanical fault diagnosis is one of the important technologies in the fourth industrial revolution. In recent years, mechanical fault diagnosis based on audio signal analysis (MFDA) has gradually become a research hotspot in the field of mechanical fault diagnosis, because of its advantages of high detection accuracy, good generalization, non-embedded measurement, and low cost. However, to the best of our knowledge, there is no comprehensive review work in this field. To facilitate colleagues to quickly understand the relevant progress in this field, this paper reviews and summarizes the research work on MFDA in recent years. First, we provide a clear definition of MFDA and introduce its development process and mechanism. Then, the related research results are classified, and the advantages and disadvantages of each method are discussed. Finally, the development prospects and challenges of MFDA are summarized. We hope that this review can provide some useful insights for researchers.