Research On Diagnosis Method Of Axle Box Bearing Based On Optimized Morphological Filter

With the rapid development of China’s high-speed railways,the number of operating trains continues to increase,and China’s high-speed trains play a pivotal role in the country.In order to ensure the safe and stable operation of the train,ensuring the normal operation of the train parts and components has become the priority of the maintenance work.Wheelset bearings of high-speed trains are different from the bearing elements on other static mechanical equipment.It not only bears the static pressure of the train,but also suffers from unsteady dynamic loads caused by radial acceleration during operation.The higher the speed,the greater the vibration and the dynamic.The greater the force,the greater the axial force when passing through the curve or in strong crosswinds.High-speed train wheelset bearings are in high-speed variable load conditions for a long time,so they are prone to failure and threaten the safe operation of trains.Moreover,the vibration signal collected at the axle box contains many frequency components and is coupled with each other,mainly including: the disturbance caused by the interaction between the wheel and the rail,the cyclic shock generated when the bearing has defect,and the irrelevant noise.How to extract effective information about bearings from disorganized signals has become a core part of fault diagnosis.In order to realize the early fault diagnosis of wheelset bearings,this thesis has carried out in-depth research on the morphological filtering algorithm.It has done in-depth research on multi-morphological filtering,demodulation effect measurement index,expansion optimization range and practical engineering application.The main research work of this thesis is as follows:(1)In this thesis,the morphological filter(MF)and multi-scale morphological filter(MMF)are introduced in detail.Firstly,the basic operation of morphological filtering is introduced and the basic operation results are explained.Secondly,the types and construction methods of structural elements are introduced.Finally,multi-scale morphological filtering is introduced to achieve multi-scale analysis using structural elements of different scales.(2)In order to improve the accuracy of data analysis,this thesis proposes an improved operation — AFBO operation based on the basic theory of morphology and verifies its feasibility through simulation signals.For the same vibration signal source,after different types of morphological filtering operation types,the demodulation effects are different and for different kinds of signals,different operation types need to be matched with them,a shape based on Cuckoo search algorithm is proposed.The learning filtering algorithm selects the operation type with the largest adaptation value and its parameter configuration as the final optimization result,and the demodulation spectrum is the optimal demodulation spectrum.(3)According to the general characteristics of the signal demodulation spectrum when the bearing is defective,the Feature energy factor is proposed as the fitness function of the cuckoo algorithm and used as a measure of the demodulation effect.(4)In order to verify the effectiveness of the above improved algorithm,the demodulation algorithm is verified by combining simulation signal,bench test and actual line test.And the demodulation effect is compared with the commonly used demodulation algorithm.Therefore,it is verified that the optimization algorithm can better recognize the cyclic shock in the signal and greatly improve the fault diagnosis capability of the wheelset bearing.