Selection principle of mathematical morphological operators in vibration signal processing

摘要

In this paper a novel method based on geometric illustration and frequency response analysis is proposed for evaluating the performance of mathematical morphological (MM) operators in vibration signal processing. With geometric illustration, the working mechanism of MM operators can be disclosed and the frequency response is studied to select suitable MM operators and structural element length. The advantage of combining geometric illustration and frequency response analysis is that it does recognize the characteristics of MM operators which are used to signal denosing or feature extraction. Selection suggestion and comprehensive explanation of MM operators for different purpose are supplied for vibration signal analysis. The paper also proposes a new operator called CMF-hat to extract the impulsive-type signal for bearing fault detection. The experimental results show that CMF-hat can effectively extract the fault features, and the proposed evaluating method advances selection of MM operators and improves the accuracy of fault diagnosis.