Study On Fault Diagnosis Of Railway Truck Rolling Bearing Based On Information Fusion

As the key component of rotating mechanical equipment,the running state of rolling bearing is directly related to the safety and stability of the equipment.It is of great significance to deeply study the characteristics of rolling bearing in different running states and establish an effective fault diagnosis system to ensure the complete and stable operation of mechanical equipment.At the same time,in the era of big data information for easier and showed a large amount of information,the information characteristics of species diversity and density of low value,through a variety of detection technology in gathering information access when the rolling bearing operation state for more isolated,and even contradictory,information fusion technology as a multidisciplinary cross technology can be effectively integrated multi-source information,can be more comprehensive and accurate characterization of the running state of rolling bearings,this study based on information fusion technology to construct trucks rolling bearing fault diagnosis system,in order to meet the railway freight inspection company demand for trucks rolling bearing automatic detection,specific work content is as follows:(1)For laboratories undertaking of China shenhua energy co.,LTD.,branch of railway freight transport truck bearing malfunction automatically detect conducting background research project,this paper expounds the significance of this study,a detailed analysis of the rolling bearing fault diagnosis at home and abroad research status and hot spots,summed up the rolling bearing fault diagnosis will be to develop in the direction of large digital and multiple information fusion,determine the design of rolling bearing fault diagnosis system based on information fusion technology.(2)In-depth study of the commonly used double row circular cone roller bearings railway wagon bearing structure,damage forms,the phenomenon and the possible causes,analyzes the bearing different damage occur often used diagnostic methods,summarizes the advantages and disadvantages of typical diagnosis methods and the suitable scenario,determine the vibration analysis method is adopted to improve the fault diagnosis of rolling bearing and the origin of rolling bearing vibration,signal characteristics,to study the vibration characteristics,analysis of its fault vibration frequency was deduced,and the important basis to judge the running state of the bearing.(3)A decision level information fusion system based on multi-angle feature extraction was proposed.This paper briefly introduces the definition,development and principle of information fusion,analyzes and compares three kinds of existing information fusion models,and proposes a new information fusion model based on multi-angle feature extraction of homologous signals and decision level fusion based on the fact that the project site is a vibration signal.(4)Involved in design and perfect the China shenhua energy co.,LTD.,branch of railway freight transport truck bearing fault automatic detection device,and the practical field of roller bearing vibration signal acquisition,using the wavelet packet transform to signal decomposition and reconstruction,and achieve the purpose of noise reduction,by comparing before and after noise spectrum,shows that the wavelet packet transform in the validity of the signal processing,and the signal after the noise reduction for effective feature extraction and bearing fault diagnosis.(5)A decision level information fusion fault diagnosis system is designed based on neural network and evidence theory,and its validity is verified by simulation.Dimensionless indexes of the signal,respectively,after the noise is spread multiscale entropy,multi-scale fuzzy entropy three kinds of characteristic index and enter the bearing running condition preliminary diagnosis RBF neural network,the output value as the input of D-S evidence theory,its diagnostic accuracy as the basic probability assignment for decision-makers information fusion,through MATLAB simulation validate the model diagnostic accuracy was 100%,significantly better than RBF neural network diagnosis system alone,shows that under the same data by multiple perspectives independent eigenvalue extraction structure multi-source body of evidence on decisionlevel information fusion method is effective.