Research On Fault Diagnosis Technology And Method Of Bearing Of Reciprocating Pump Based On Uncertainty Theory

Reciprocating pump is a widely used fluid machinery in the petroleum and petrochemical industry.It is often used to transport high-pressure liquid with high viscosity,high density and high sand content,with relatively slow flow rate.Reciprocating pumps commonly used in oil mines include oil pumps,drilling pumps,cementing pumps and fracturing pumps.As the most critical and most vulnerable parts of reciprocating pumps,bearings need more excellent quality to meet the increasing harsh working environment.The dissertation focused on the fault diagnosis techniques of rolling bearing,the key components of three-cylinder single-acting horizontal reciprocating piston pump.And the types of rolling bearing studied include drive shaft bearing,passive shaft bearing,connecting rod bearing and crosshead bearing.Aiming at the non-stationary and non-linear vibration signal of the dynamic end rolling bearing of reciprocating pump and containing a large amount of noise,a fault diagnosis method based on EEMD,distance factor,correlation coefficient and wavelet packet decomposition is proposed.By measuring the bearing vibration signal on the bearing life test bench,the measured signal is decomposed by EEMD,and the IMF components are selected and reconstructed according to the distance factor and the correlation coefficient to highlight the fault feature information and avoid the influence of accidental error.The weight The vibration signal is decomposed by wavelet packet to construct the energy signature signal vector,then the correlation coefficient of the eigenvector is calculated,and the fault type is judged by comparing the absolute value of the difference of the correlation coefficient.The method of fault recognition rate and the bearing vibration signal correlation coefficient analysis compared to the fault recognition rate has greatly improved,and without the need for neural network identification requires a lot of data training,is a better method of bearing fault identification.Aiming at the non-stationary and nonlinear vibration signals of the reciprocating pump dynamic end rolling bearing,this paper presents the general EEMD algorithm,the ApEn and the SVM combined rolling bearingFault diagnosis method.The method obtains the IMF component by measuring the bearing vibration signal and using EEMD to decompose the measured signal.Based on the study of several commonly used IMF components,a screening method based on approximate entropy The IMF extracts the energy value and inputs it directly into the SVM to establish a fault prediction model.At the same time,the extracted IMF is extracted and multiplied by its corresponding approximate entropy as a new eigenvalue,which is used as the SVM parameter input to establish another fault prediction model.By contrast,the latter effect is better,that is,diagnostic efficiency is higher.The original signals of rolling bearing collected through the test often contains noise information,redundant information,and other interference information,and can not directly diagnosis the malfunction of the rolling bearing.To solve this problem,we analyzed and processed the original signals and extracted the characteristic parameters of the rolling bearing to determine its fault status.These characteristic parameters include time domain feature parameters,frequency domain feature parameters,and time-frequency domain feature parameters.The malfunction of the reciprocating pump rolling bearing has randomness,and is affected by the sensor measurement error and the complexity of the environment,resulting in the randomness,uncertainty and incompleteness of the failure information obtained from the test.Therefore,this paper adopted a combination of cloud model and D-S evidence theory by screening and removing the uncertain factors and merging eigenvalues to eliminate the uncertainties,and hoped to achieve more accurate fault diagnosis for rolling bearings.Through the above study,we have come up with a complete fault diagnosis method in this paper.The method achieves accurate fault diagnosis from experimental design,acquisition of the original vibration signals,and extraction-selection-fusion of characteristic parameters.In summary,this study has a certain role in the development of fault diagnosis methods and techniques for the rolling bearings.