Fault Diagnosis Of Rolling Bearing In Large Petrochemical Units Under Strong Environment

The potential safety hazard of petrochemical equipment will not only affect its normal production,but also may result in serious safety accidents,doing harm to the environment as well as making losses in economy.Large petrochemical machines,with complicated compositions,are well susceptible to strong interference,as a result of working in a hostile environment for a long time.The antifriction bearings,as the important part of large petrochemical machines,its working situation influences the operation of the whole equipment.Bearing breakdown are generally classified of inner race faults,outer race faults and rolling element faults and so on.Based on the above problems,with the aim of ensuring safe,stable,long--period,full load and optimized operation and production of the petrochemical equipment,it is well necessary to diagnose the antifriction bearing faults of large petrochemical machines with strong interference.As long as the vibration signal analysis by the traditional EMD and the scanty extraction of fault characteristics by the wavelet as well as deficiency diagnosis classification of fault data by the traditional support vector machine(SVM),this passage has investigated the improvement of the EMD algorithm and the extraction of the wavelet packet features as well as recognition and classification of bearing faults by the optimized SVM,according to rolling bearing vibration signal feature of petrochemical large machines under the situation of strong interference.(1)According to the fundamental compositions and working principle of the rolling bearing,it has analyzed the normal bearing and three common bearing failure forms(outer race fault,outer race faults and rolling element faults)as well as its cause and necessary measures.Then,through introducing the vibration mechanism of the bearing,it makes sense of that the bearing will be disturbed by internal and external elements when operating.(2)Directing at the limitations of analyzing and applying the bearing vibration signals by the traditional empirical mode decomposition(EMD),it has made an excellent response,through analyzing and applying the simulation signals as well as actual rolling bearing vibration signals with the optimized EMD.(3)Directing at the limitations of the feature extraction towards the four states data of rolling bearing by wavelet algorithm,it extracts the four states features of bearing vibration data through wavelet packet analysis.(4)According to the limitations of discerning and classifying the four states by traditional SVM,it puts forward that it has achieved the ideal outcomes with the combination between the optimized support vector machine and the eigenvector of four states data extracted by wavelet packet algorithm as well as the identification and classification of its fault features.