SVD Theory for Machine Fault Detection: A Review

Due to the complex working environment and working conditions of mechanical equipment, its key components are easily damaged, resulting in a decline in equipment performance. Condition monitoring and fault diagnosis are considered to be necessary means to ensure the safe service of mechanical equipment. Singular value decomposition (SVD) theory can effectively suppress the interferences of noise and other irrelevant signal components in the signal of mechanical equipment, to achieve the effective extraction of fault features. Due to the unique properties of SVD, it has been widely applied in the domains of signal processing and fault detection. At present, scholars have carried out a mass of research work on SVD and its applications in mechanical fault diagnosis, as well as other fields. This article expounds on the theoretical development of SVD and attempts to summarize and generalize the theoretical progress of SVD and its application in mechanical fault detection and other fields from signal noise reduction, signal separation, and other applications. The problems and challenges faced by SVD theory and its applications in mechanical fault diagnosis are analyzed, to provide comprehensive and valuable guidance.