New Dimensionless Parameter Construction And Experimental Verification In Complicated Environment

Equipment safety is the basis of safety,stability,long cycle,full load,and optimized production operation of petrochemical industry.Reducing the failure rate and improving the reliability are the most important issues of petrochemical enterprises at home and abroad.With the increasingly large scale,integration and complexity of petrochemical equipment,petrochemical equipment safety is facing greater challenges.The position of large rotating machinery in petrochemical equipment is very important.The main processes such as catalysis cracking can not be separated from rotating machinery,but in the fault bank of rotating machinery,the bearing fault accounts for about 44% of the total number of faults.Therefore,it is significant to diagnose rolling bearings in rotating machinery.Professor Zhang Qinghua's previous research results show that dimensionless immune diagnosis technology has good potential and diagnostic advantages in bearing fault diagnosis.On the basis of previous studies,aiming at the problem of rolling bearing fault diagnosis in complex environment,a new dimensionless index construction method and diagnostic technology is proposed in this paper.Specific work includes :(1)The fault diagnosis of rolling bearings in rotating machinery is difficult to solve in the traditional dimensionless index.The fault mechanism of rolling bearing is analyzed.The vibration signal of bearing normal operation is "standard vibration signal",and the concept of "mutual dimensionless index" is put forward based on the traditional dimensionless index,a new dimensionless index is constructed by normalization.(2)A wavelet packet denoising preprocessing method is proposed.In order to reflect the real vibration of the machine more accurately,we can preprocess the vibration signal in complex environment.In this paper,a preprocessing method of wavelet packet denoising is proposed.The simulation results show that the method can effectively improve the signal to noise ratio and extract the original feature of the signal.(3)A new non dimensional index fusion diagnosis method based on random forest model is designed.By combining the wavelet packet denoising algorithm,the dimensionless index and the random forest algorithm,the framework of preprocessing,model building,diagnosis mechanism and parameter selection have been put forward,and a new method of systematic,convergent and low time complexity is formed.The simulation and the verification of the semi physical unit show that the method has good recognition ability for the bearing fault,and the diagnosis accuracy can be greatly improved compared with the traditional dimensionless index method.Based on the Key Laboratory of Guangdong petrochemical equipment fault diagnosis,the work contents are simulated and verified by semi physical unit.The result of this paper is a powerful supplement to the non dimensional immunodiagnostic technique,and a new exploration of the fault diagnosis technology for the key parts of the rotating machinery in the petrochemical equip ment rotating machinery.