Cracks Recognition Of Pantograph Based On Laplace Sparse Decomposition Modal Parameter Extraction

The majority of rail transit vehicles in China adopt electric traction,and a widely used power system is pantograph-catenary,which set pantograph on the top of vehicle to obtain power from catenary by sliding contact.With the increase of the vehicle speed,the reliability and stability of pantograph-catenary system are facing more and more challenges under the various impacts from uneven elasticity of catenary,wheel-track system through car body and air disturbance,etc.In actual operation,various kinds of pantograph failures occur from time to time,which may lead to the delay of the vehicle or even affect the whole railway system.Therefore,research on the fault diagnosis for the pantograph system is of great significance to ensure the normal operation of railway system.Although China has built the high-speed railway power supply safety monitoring system(as know as 6C system)in 2012,however,the diagnosis function of the system is relatively limited,and due to the limitation of shooting angle,the identification of pantograph structural crack cannot be realized.Therefore,the performance evaluation,fault diagnosis and on-line detection technology for pantograph system are still in its infancy,which is far from meeting the growing demand of vehicle operation.In this dissertation,the fault diagnosis methods were developed to detect the structural cracks of pantograph in following contents:(1)The pantograph fault mode and occurrence frequency,including mechanical components faults,electrical components faults and gas path components faults,were systematically investigated.The high proportion of pantograph head structural cracks were concerned as the research object of this dissertation and carry out the corresponding fault diagnosis research.Firstly,the relationship between the structural crack and the vibration characteristics of the pantograph were qualitatively analyzed.By constructing the finite element model of the pantograph,the vibration characteristics of the pantograph were simulated under the influence of the fault forms of different degrees of pantograph horn crack and pantograph head support crack.It was found that these fault forms and degrees can be characterized by the modal parameters of the pantograph,the natural frequency and damping ratio can be used as characteristic parameters to identify the structural cracks of pantograph head,which provides theoretical support for the crack fault characteristics extraction of pantograph structure.(2)The non-stationary environmental excitation in the actual operation of pantograph were analyzed,and the response signal processing method based on non-stationary environmental excitation was developed.An improved random decrement technique(RDT)algorithm based on empirical mode decomposition(EMD)was proposed to overcome the adverse effect of response signal trend term on the extraction results.The approximate free attenuation signal extracted by the above method was used for the extraction of modal parameters extraction.(3)The idea of using sparse decomposition algorithm was proposed to deal with the problem of extracting modal parameters from approximate free attenuation simulation signals.The basis functions and sparse decomposition algorithms suitable for pantograph vibration signals were studied.The simulation results showed that the proposed sparse decomposition method based on Laplace basis is effective in processing approximately free attenuated signals.Meanwhile,further discussion was carried out for the situation that the initial phase of the measured signal is none-zero,which completed the deficiency of the traditional Laplace basis lack of initial phase parameters and built a solid foundation for the further engineering application of the method.(4)The pantograph rig test and artificial crack fault were designed.Based on the measured data from the test rig,the effectiveness of the Laplace based sparse decomposition method was verified for pantograph crack fault diagnosis.For overcoming the shortage of traditional sparse decomposition that requires setting dictionaries according to experience and constructing Laplace-based redundant dictionaries,a sparse decomposition optimization algorithm based on Gaussian scale space was proposed to effectively improve the operation efficiency.Finally,the classification results based on random forest classifier showed that the proposed method in this dissertation can effectively classify the location and degree of pantograph failure,which justify the effectiveness of the proposed algorithm.