Bearing Fault Diagnosis Of Mine Reduced Gear Based On BOA-SVM

In the large mechanical and electrical equipment of mining enterprises,the reducer for mining plays the key role of power transmission in the production process,and the normal operation of the reducer is of great significance to the safety production of mining enterprises.As a key component of mine reducer,rolling bearing is easy to be damaged because it often runs under the condition of large load,continuous long time operation or not timely maintenance.Rolling bearing faults not only affect the normal work of the reducer,causing unplanned shutdown of the enterprise,but also may cause downtime of underground ventilation,water supply and drainage,coal and gangue transportation and other links due to mechanical and electrical equipment faults,leading to safety production accidents.In serious cases,it will even bring threats to the personal safety of underground workers.Therefore,it is of great practical significance to study and establish the bearing fault diagnosis system of mine reducer for reducing accidents and ensuring the smooth operation of equipment.In this dissertation,the fault types of mining gear reducer bearings are studied deeply,and a hardware platform of signal acquisition is designed to realize effective signal acquisition under complex working conditions.The signal acquisition platform can collect the vibration data of the rolling bearing of the reducer at high speed,and upload the monitoring data to the ground industrial computer through Ethernet or RS485 interfaces.According to the characteristics of rolling bearing fault data,the wavelet packet transform technology was used to extract the reconstructed energy values after vibration signal decomposition,and the feature vectors were constructed by using the energy value ratio of each frequency band.The Support Vector Machine(SVM)classification model was used to identify the fault features of rolling bearings.Since the classification effect of SVM depends on the selection of key parameters,this dissertation proposes to use Bayesian Optimization Algorithm(BOA)to find the optimal parameter value of SVM.A fault diagnosis model based on wavelet packet decomposition energy algorithm feature extraction and BOA-SVM was established.In order to verify the effectiveness of the proposed fault diagnosis method,besides Bayesian optimization algorithm,Particle Swarm Optimization(PSO)and Genetic Algorithm(GA)are used to optimize the key parameters of SVM.By comparing the experimental results of PSO-SVM and GA-SVM classification models,it is found that the BOA-SVM classification model can diagnose bearing faults more quickly and effectively improve the accuracy of fault diagnosis.It shows that this dissertation has a certain theoretical significance and practical value in the field of fault analysis of rolling bearing and fault diagnosis of mine reducer.