A Research On Rail Fastening State Detection Based On MEMS And Information Entropy

As a crucial component of the track on a railway,a rail fastening system is used to bind rails and the sub-rail foundation together and provide elasticity.With the development of high-speed and heavy-haul railways,the rail structure is suffering from increasing loads at a high speed and it is possible for the fasteners to become loose or even fail due to the destructive vibration produced by severe impact of train wheels.And this will lead to various defects and damages on the surface of track structures,or even cause serious harm to the safety of rail transit.However,traditional railway inspection methods are implemented in manual ways that are inefficient and high cost,and often with subjective results dependent on the human inspectors.The status of the artificial detection of rail fastener has been unable to complete the heavy rail maintenance tasks and can’t conform to the development trend of the high-speed railway technology in China,so a rapid,efficient,accurate and low cost of rail fastener condition monitoring technology solution is needed.In this context,this paper designed a railway fastener status detection system,which can monitor the loosening status of fasteners in real time and upload the results to the cloud server,so that the relevant railway maintenance departments can make timely response.The specific work and corresponding research results are as follows:This paper designed and developed a passive wireless and low-cost fastener detection system,which consists of a low-power Micro Inertial Measurement Unit(MIMU)and a Global System for Mobile Communications(GSM)unit.The system is powered by light energy without external power supply.The acceleration signal measured by the sensor is uploaded to the cloud server and the status information of the rail fasteners is obtained after data processing.Aiming at the mechanical structure and connection characteristics of railway fasteners,the dynamic model and finite element model of the system are established in this paper,and the theoretical analysis was used to analyze rail vibration characteristics in the vertical direction excited by pulse signals.On this basis,the Power Spectrum Entropy theory was applied to identify fastener looseness reliable.In addition,the simulation results show that the information entropy is not disturbed by the singularity to a large extent,which is beneficial to reduce the false alarm of the fastener state detection system and improve the practical ability.Field experiment results show that when the fastener torque changes in the range of 60 N· m-140 N·m,the power spectrum entropy changes little,but for fasteners with a large degree of looseness,accurate identification can still be achieved.In view of the deficiency of information entropy method in fastener state recognition,a signal feature extraction method based on IMF distributed entropy was proposed according to the theoretical knowledge of signal feature extraction.The signal is first antisymmetric extended,and then decomposed by EMD method.The distribution entropy of each IMF signal in the time domain was obtained using amplitude entropy method,which characterize the characteristics of the signal in different frequency bands.Then the distribution entropy is used as the classification feature,and the SVM classifier is used to get the looseness level of fasteners.The classification results show that compared with the results using amplitude entropy alone,this method can improve the identification effect of fastener state greatly.The railway fastener condition detection system based on IMU and GSM proposed in this paper can accurately identify the shedding and loosening state of fasteners,which has low cost,and can bring significant benefits for the day-to-day maintenance of railways.