Knock detection based on the optimized variational mode decomposition

• Published in: Measurement : journal of the International Measurement Confederation 2019-07, Vol.140, p.1-13
• Main Authors: Bi, Fengrong, Li, Xin, Liu, Chunchao, Tian, Congfeng, Ma, Teng, Yang, Xiao
• Format: Article
• Language: English
• Subjects: Adaptive decomposition; Algorithms; Decomposition; Engine; Engines; Fault diagnosis; Knock; Optimization; Recursive methods; Signal processing; Variational Mode Decomposition (VMD); Vibration
• ISSN: 0263-2241; 1873-412X
• DOI: 10.1016/j.measurement.2019.03.042

•VMD is introduced to decompose vibration signals to detect knock precisely.•Combined with recursive model, an optimized recursive VMD is proposed.•The customized initialization of center frequencies ensure efficiency and stability.•A stop criterion based on energy difference tracking method is proposed.•Results show the optimized VMD has advantageous in accuracy and efficiency. In order to extract slight characteristics and achieve exact control of engine knock to improve the overall performance of engine, the optimized Variational Mode Decomposition (VMD) algorithm is introduced to process the vibration signals of engine. Combined with recursive model and energy difference tracking method, the optimized VMD can decompose signals into several modes based on stop criterion instead of artificial values set in advance. Besides, the initial values of center frequencies are also optimized for stability and efficiency. An analog signal is firstly decomposed by this method, and result shows that the optimized VMD has ability to extract components accurately. Then this method is applied for the decomposition of tested engine vibration signals and compared with the Empirical Mode Decomposition (EMD) and traditional VMD, the results show that the optimized VMD can eliminate effect of noise and extract knock at different intensities more accurately and quickly.