Study On Optimal Sensor Placement Methods Of High-speed EMU's Bogie Frame

With the development of Chinese high-speed railway,the speed of railway is increasing,and the requirement for safety of railway vehicles is becoming higher and higher.As the main receiving component of the bogie,frame is the key determinant of safe operation.Structural health monitoring plays a critical role in ensuring safety,developing the technology of the high-speed railway,and collecting information by optimal sensor placement which is the prerequisite for the effective application of structural health monitoring technology.In this thesis,the reliability of frame is verified by the international railway alliance standard,and determining the dangerous position of the frame which is the position of the maximum stress value under various working conditions.Because the optimal sensor placement is vulnerable to the mode number during vibration modal analysis,this thesis proposes entropy-based approach to determining the optimal number of target modes.Then,optimizing the location and quantity of the sensor placement are based on the Fisher information criterion,frequency response function,independent component analysis,K-clustering method and information entropy.The research contents and main conclusions are as follows:(1)For establishing the finite element model of the frame,ensuring the reliability of the frame and determining the dangerous position of the frame,firstly,using Solidworks modeling software to solid modeling and simplifying the model according to the relevant of CRH3's bogie frame is the first step.The finite element model of the frame is established by ANSYS Workbench finite element analysis software.Then,frame does the Static strength analysis based on UIC615-4 standard.Analysis results show that,the static strength of the frame meets the strength requirement and determine the dangerous position of the frame.(2)In order to ensure the mode number of model analysis and to certify the reasonable result of sensor placement,this thesis proposes an entropy-based approach to determining the optimal number of target modes.Firstly,compute the entropy in various stages according to the Fisher information matrix,the sudden change characteristic function of the entropy is computed to determine the sudden change point.In terms of the change of entropy,which implies the change of the quantity of Fisher information matrix,the mode numbers of global and local vibrations can be calculated.Simulation results of luggage rack and CRH3 bogie frame show that this new approach can determine the number of the global vibration on the frame precisely and extract the first seven global vibration modes.(3)In order to realize both location and quantity optimization,a method of optimal sensor placement based on frequency response function is proposed.Firstly,the structure modal analysis is conducted and mode shapes are extracted,and the frequency response function of the structure is calculated.Then,based on independent component analysis and K-clustering method,the sensors' positions are optimized.Finally,the Fisher information matrices and their entropies corresponding to the placement results of different sensor numbers are calculated.The number of sensors that shows the minimum figure is optimal.The simulation results of CRH3 bogie frame show that the sensor placement based on frequency response function can achieve the optimal arrangement of sensors which are places nearby the maximum stress positions from every condition based on UIC615-4 standard,and can maximize the information which is obtained by the limited sensors.