Feature extraction method of bearing fault signal based on adaptive multi-scale avg-hat transformation

Abstract

The invention discloses a bearing fault signal feature extraction method based on adaptive multi-scale AVGH transformation. The steps are as follows: 1. According to the parameter index of the bearing fault signal, determine the number of initial multi-scale structural elements and the initial structural elements of the signal 2. Construct a set of initial multi-scale structural elements; 3. Calculate the result of the corresponding morphological AVG-Hat transformation of the bearing fault vibration signal under the initial multi-scale structural elements, and construct a set of the results; 4. Through particles In the group optimization method, the ratio of the permutation entropy of the filtered bearing fault vibration signal to the sparsity of the envelope spectrum is selected as the evaluation index, and the optimal weight coefficient corresponding to the initial multi-scale structural element after filtering is adaptively determined; 5. According to the weight coefficient construction Optimal multi-scale morphological AVG‑Hat filter; 6. Calculate the result of the filter processing the vibration signal of the bearing fault, and extract the fault characteristic components in the signal through the envelope spectrum analysis of the signal to diagnose the bearing fault.

Description

technical field [0001] The invention relates to a bearing fault signal feature extraction method based on adaptive multi-scale AVG-Hat transformation, which belongs to the technical field of mechanical fault diagnosis and signal processing. Background technique [0002] In actual engineering, the vibration signal of rolling bearing fault is a typical nonlinear and non-stationary signal, and the fault characteristics in the signal are easily covered by various background noises. Therefore, it is very difficult to diagnose bearing faults under strong background noise. Mathematical morphology is a typical nonlinear signal processing method. It uses specific scale and shape structural elements to fit and modify the local details of the signal time domain waveform. It can effectively eliminate the background while extracting the main waveform features in the signal. noise interference. The key to using morphological methods to process fault signals is to construct a morphologica...

Images