A new adaptive evolutionary digital filter based on alternately evolutionary rules for fault detection of gear tooth spalling. (1st March 2019)

• Record Type: Journal Article
• Authors: Wang, Liming; Ye, Weijun; Shao, Yimin; Xiao, Huifang
• Is Part Of: Mechanical systems and signal processing. Volume 118(2019)
• Page Start: 645
• Page End: 657
• Publication Date: 2019-03-01
• Subjects: Fault diagnosis -- Gear tooth spalling -- Evolutionary digital filter (EDF) -- Adaptive noise cancellation (ANC); Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2018.09.005
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: A new adaptive evolutionary digital filter is proposed for gear fault detection. The speed and noise cancellation performance of the proposed method are validated. The proposed method exhibits much faster convergence rate than other methods. Abstract: Evolutionary digital filter (EDF) is an adaptive filter that controlled by adaptive algorithm based on the evolutionary strategies: cloning and mating. The adaptive algorithm of EDF exhibits the advantage of avoiding local optimum problems that arise from a multiple-peak performance surface of an adaptive noise cancellation (ANC) system. However, the convergence rate of the classic EDF is limited by inappropriate arrangement of evolutionary strategies and large population of individuals. In this paper, a new method referred to alternately evolutionary digital filter (AEDF) is proposed in order to solve the problem of lower convergence rate of the classic EDF. The basic concept of the proposed method is to alternately apply mating evolutionary strategy to locate the peaks in performance surface quickly and then apply cloning evolutionary strategy to get the global optimum near the peaks accurately. The convergence rate and noise cancellation performance of the proposed method is verified by both simulated signals and experimental signals. The results show that the proposed AEDF-based ANC method can effectively detect the gear tooth spalling faults in noisy environment and exhibits much faster convergence rate compared to previous methods.