Bearing fault intelligent diagnosis method based on compressed sensing and correlation vector machine

Abstract

The invention discloses a bearing fault intelligent diagnosis method based on compressed sensing and a correlation vector machine. According to the method, fault diagnosis is achieved through vibration signal analysis. The method comprises the steps of firstly selecting a Gaussian random matrix as a measurement matrix based on a compressed sensing theory to realize compressed sampling of signals,secondly constructing an over-complete redundant dictionary to perform sparse representation on the signals, then utilizing an orthogonal matching pursuit algorithm to realize signal reconstruction, and selecting a time domain index sensitive to fault features as a feature vector for the reconstructed signals; and finally, selecting a Gaussian function as a kernel function, dividing a training sample and a test sample by utilizing a feature vector, importing the training sample into an intelligent recognizer of a relevance vector machine model constructed by a relevance vector machine, and comparing a test result with an actual fault type and degree to obtain the effectiveness of the diagnosis model. According to the invention, the problems of difficult transmission and processing of massdata and low signal-to-noise ratio of bearing vibration signals can be solved, and qualitative and quantitative identification of bearing faults can be realized more accurately.

Description

technical field

[0001] The invention belongs to the technical field of fault diagnosis, and in particular relates to an intelligent diagnosis method for bearing faults. Background technique

[0002] Bearings are the core components of mechanical equipment, and they are also one of the most vulnerable components. The operation of rolling bearings in large rotating machinery is directly related to the safe and economical operation of the equipment. Once the rolling bearing fails, it will cause a chain reaction. Cause the transmission system to fail, and seriously cause the serious accident of machine crash. Therefore, it is of great significance to monitor the running status and fault diagnosis of rolling bearings.

[0003] However, the traditional real-time monitoring of bearing vibration signals will inevitably generate massive data. At the same time, the acquisition of vibration signals needs to meet the requirements of the Nyquist sampling theorem, resulting in difficulti...

Examples

specific Embodiment

[0084] In this embodiment, the bearing test bench is connected with a data acquisition instrument to collect vibration signals of bearings in different working states. During the experiment, the sampling frequency is set to 15360Hz, the number of sampling points is 8192, and the motor speed n=1496r / min. Bearing parameters are as follows: outer diameter D=80mm, inner diameter d=35mm, number of rolling elements Z=8, contact angle α=0°. In the pitting test of the bearing, the pitting on each component of the bearing is a single point pitting, and the pitting defects are small pits with a diameter of 2mm and a depth of 0.1mm. The process of fault intelligent identification using the correlation vector machine model is as follows:

[0085] (1) First, install the acceleration sensor in the vertical direction of the bearing seat to collect the mechanical vibration signal of the bearing x N×1 , where N is the number of sampling points.

[0086] (2) Selecting a measurement matrix to...

technical field

[0001] The invention belongs to the technical field of fault diagnosis, and in particular relates to an intelligent diagnosis method for bearing faults. Background technique

[0002] Bearings are the core components of mechanical equipment, and they are also one of the most vulnerable components. The operation of rolling bearings in large rotating machinery is directly related to the safe and economical operation of the equipment. Once the rolling bearing fails, it will cause a chain reaction. Cause the transmission system to fail, and seriously cause the serious accident of machine crash. Therefore, it is of great significance to monitor the running status and fault diagnosis of rolling bearings.

[0003] However, the traditional real-time monitoring of bearing vibration signals will inevitably generate massive data. At the same time, the acquisition of vibration signals needs to meet the requirements of the Nyquist sampling theorem, resulting in difficulti...

Examples

specific Embodiment

[0084] In this embodiment, the bearing test bench is connected with a data acquisition instrument to collect vibration signals of bearings in different working states. During the experiment, the sampling frequency is set to 15360Hz, the number of sampling points is 8192, and the motor speed n=1496r / min. Bearing parameters are as follows: outer diameter D=80mm, inner diameter d=35mm, number of rolling elements Z=8, contact angle α=0°. In the pitting test of the bearing, the pitting on each component of the bearing is a single point pitting, and the pitting defects are small pits with a diameter of 2mm and a depth of 0.1mm. The process of fault intelligent identification using the correlation vector machine model is as follows:

[0085] (1) First, install the acceleration sensor in the vertical direction of the bearing seat to collect the mechanical vibration signal of the bearing x N×1 , where N is the number of sampling points.

[0086] (2) Selecting a measurement matrix to...

Images