Extraction of mechanical multi-fault using parameter-adaptive jump plus mode decomposition

• Published in: Signal, image and video processing Vol. 19; no. 10; p. 821
• Main Authors: Han, Yue, Yang, Na, Wei, Yu, Xu, Yuanbo
• Format: Journal Article
• Language: English
• Published: London Springer London 01.10.2025; Springer Nature B.V
• Subjects: Algorithms; Bandwidths; Computer Imaging; Computer Science; Decomposition; Fault detection; Fault diagnosis; Faults; Image Processing and Computer Vision; Mechanical systems; Methods; Multimedia Information Systems; Optimization; Original Paper; Parameters; Pattern Recognition and Graphics; Search algorithms; Signal,Image and Speech Processing; Sparsity; Vision; Wavelet transforms; Feature extraction; Multi-fault; Improved sparrow search algorithm; Adaptive parameter selection; Jump plus mode decomposition
• ISSN: 1863-1703; 1863-1711
• DOI: 10.1007/s11760-025-04325-y

Given the intricate nature and challenging operational conditions experienced by mechanical systems, it is common for multiple faults to simultaneously occur within machines. This means that different types of faults can be coupled together. One major challenge faced when extracting multi-fault is distinguishing between weak and strong fault characteristics. Moreover, the fault signals are heavily distorted due to the presence of strong ambient interference, such as noise and vibrations generated by other mechanical interactions within the system. To overcome this challenge, a novel multiple faults extraction technique based on a parameter-adaptive jump plus mode decomposition (PA-JPMD) based on an improved sparrow search algorithm is proposed in this paper. The PA-JPMD can adaptively find the optimal predefined parameters through the characteristics of signals, thus avoiding the blind selection of the parameters. Furthermore, the PA-JPMD is capable of decomposing multi-fault signals into high and low-frequency components based on the oscillation characteristics associated with different types of faults, thereby identifying all potential fault features. The validity and practicality of the PA-JPMD are confirmed through the utilization of two groups of real experimental data from two different test rigs. The PA-JPMD demonstrates superior capability in identifying weak multi-fault in the presence of ambient noise, as evidenced by comparisons with existing similar methods.