A Separation Method of Continuous Rolling Mill Vibration Signals Based on Sparse Feature Similarity

Abstract

The invention discloses a continuous rolling mill vibration signal separation method based on sparse feature similarity. The method is as follows: Acceleration sensors are arranged at one or more different positions on the surface of the sensitive part of the rotating part of the continuous rolling mill, and the vibration signals when each vibration source is working are collected, that is, the observation signals; through the sparse feature extraction method based on time-spectrum segmentation, the vibration signals of each channel are obtained. Sparse representation of the collected vibration signals under the optimal set of atoms; then cluster all the atoms extracted from the sparse representation through structural similarity, estimate the number of vibration sources, and use the atoms corresponding to each cluster center to update the sparse representation Results; Finally, the mixing matrix is ​​estimated through the clustering coefficient corresponding to the new sparse representation result, and then the independent signals of each vibration source are calculated to realize the separation of the vibration signals of the continuous rolling mill. This method can effectively filter out the interference noise during the operation of the continuous rolling mill, improve the accuracy of sparse feature extraction, and estimate the number of vibration sources and the mixing matrix more accurately.

Description

technical field [0001] The invention belongs to the technical field of state monitoring and fault diagnosis of rotating machinery, and relates to a method for separating vibration signals of a multi-stand continuous rolling mill, in particular to a method for separating vibration signals of a continuous rolling mill based on sparse feature similarity. Background technique [0002] In the strip production process of metallurgical enterprises, the rough rolling unit and the finishing rolling unit are the core equipment of the entire rolling process. The working status of the bearings and rolls in these continuous rolling units will directly determine the quality accuracy and performance indicators of the rolled products. During the long-term continuous and high-intensity work of the continuous rolling mill unit, failures such as cracks on the surface of the roll, partial peeling of the surface of the roll, and failure of the support bearing of the roll often occur, which will a...

AI Technical Summary

Problems solved by technology

However, there is a defect in the above-mentioned patent: in engineering applications, there are often cross-frequency components between mechanical vibration sources, which makes it impossible to satisfy the independence assumption of the independent component analysis algorithm, thus causing the algorithm to fail

However, this method has shortcomings: when fitting the clustering lines of each vibration source through the time-frequency scatter diagram and then calculating the mixing matrix, the estimation accuracy and robustness of the direction of the line are often poor due to noise interference , a good estimate of the mixing matrix cannot be reached

Images