Fault Diagnosis Of Rolling Bearing Based On Graham Angle Field Method Study

With the development of modern manufacturing and the advancement of science and technology,electromechanical equipment is developing in the direction of integration,precision,high speed and intelligence.As a key component of rotating machinery,rolling bearings are of great significance for real-time monitoring of bearing operating status and fault diagnosis.This paper takes the vibration signal of rolling bearing as the research object,analyzes various methods of bearing fault diagnosis,and summarizes two difficulties encountered in bearing fault diagnosis:(1)When the one-dimensional bearing vibration signal is used as the input of the machine learning model,the time sensitivity is poor and the feature extraction is slow.(2)It is difficult to express and extract the characteristics of the vibration signal under the influence of noise and variable working conditions.Facing the above two problems,this paper proposes two rolling bearing fault diagnosis models based on the Graham angle field(GAF):(1)A bearing fault diagnosis model that converts vibration signals into GAF images is proposed.The collected vibration signals of rolling bearings are converted into GAF feature maps by using the Graham angle field(GAF)image encoding method,and more geometric properties and intrinsic data structures are displayed through the high-order signal description capabilities of GAF images.The time characteristics of the vibration signal and the relationship between the signal sequences are directly expressed in the image.Solve the problem of efficiency when the one-dimensional vibration signal is used as the input of the machine learning model,and the difficulty of feature expression and extraction under variable working conditions.Afterwards,the GAF image is input into the residual network as a data set,so that the neural network can further extract deep features from the texture information of the GAF image,and complete the extraction and classification of ten different fault types of rolling bearings.(2)A bearing fault diagnosis model that converts frequency domain information of vibration signals into GAF images is proposed.Aiming at the problem that the accuracy of the model drops seriously under high noise when the time-domain signal is directly converted into the GAF feature map.The model converts the bearing vibration signal into a frequency domain signal and encodes it into a GAF image after preprocessing,which improves the influence of the noise problem in the feature extraction of the model.At the same time,in view of the unique spatial distribution characteristics of the Graham angle field method,it is proposed to add channel attention mechanism(ECA)to the improved residual neural network,which further improves the anti-noise performance of the model.Finally,the performance of the model is analyzed under different noises and different working conditions,and comparative experiments are carried out with other algorithm models.The experimental results show that the improved GAF bearing fault diagnosis model based on frequency domain transform proposed in this chapter has good anti-noise performance.