Rolling Bearing Fault Diagnosis Based on Mathematical Morphological Spectrum

Mathematical morphology is a type of signal processing that is often used in image feature extraction. This paper fully explains the efficiency of extracting rolling bearing fault information by using hit-miss transformation and morphological spectral features during mechanical fault signal feature extraction. Mathematical morphological spectra can use multi-scale structural elements for morphological feature extraction, but for one-dimensional signals, the translational invariance and spatial insensitivity of the morphological spectrum will lead to the problem that the overall feature extraction difference of the feature under strong noise is not obvious. This paper proposes a problem that is not considered in this field, and combines the morphological spectrum and hit-miss transformation to efficiently extract the characteristics of the overall and fault pulse signals, which reduces the classification time and improves the recognition efficiency.