| The safety and comfort of the train are closely related to the running quality and health state of the bearing.At present,the railway department mainly adopts the method of periodic repair,which makes some bearings with long life not fully utilized.If the bearing fault is found in time and measures are taken to repair them in operation,the service cycle of the bearing can be extended to the maximum extent and the maintenance cost of high-speed railway can be reduced.In this paper,the rolling bearing fault in high-speed railway axle box is the main research object.Firstly,the structure and vibration mechanism of the bearing are discussed.According to the structure and operating characteristics of the bearing,the types and causes of the faults that often occur in the bearing are summarized,and the characteristic frequencies of different faults of the bearing are calculated.The effect of several signal decomposition methods on the decomposition of the faulty bearing vibration signal is studied,and the advantage of the Variational mode decomposition(VMD)in signal decomposition is focused on.Propose the KH-VMD algorithm that uses Krill herd algorithm(KH)to optimize VMD,and prove the reliability of the optimization.Secondly,the relationship between various types of fault information and entropy value of rolling bearing is explored,and the result shows that the entropy-fuzzy entropy is the most suitable for characterizing fault types.The KH-VMD algorithm is used to decompose the signal into multiple dimensions,and the fuzzy entropy value of the signal in each dimension is extracted.Take it as the characteristic quantity of bearing failure,and prove the validity of this characteristic quantity.Finally,based on cuckoo search algorithm(CS)and extreme learning machine(ELM),CS-ELM algorithm is proposed for fault identification and classification.Based on the KH-VMD algorithm,the fuzzy entropy and the CS-ELM algorithm,the KHVMD-FE-CSELM algorithm is proposed to achieve the goal of directly determining the type of fault from the vibration signal.The KHVMD-FE-CSELM algorithm is tested through experiments to collect vibration signals of different faults.The results show that this algorithm has a higher accuracy rate than other automatic identification algorithms.In summary,the KHVMD-FE-CSELM algorithm proposed in this article has a certain reference value for the research on the fault identification of rolling bearings.It makes up for the misjudgment that may be caused by insufficient manual experience,and has certain practicability. |
