Research On Intelligent Sensing Algorithm For CA Mortar Disengagement Of Ballastless Track

Ballastless track has been widely used with its advantages including high stability,good stiffness uniformity,strong structure durability,and less maintenance requirement.As an important part of slab ballastless track,Cement Asphalt Mortar(CA Mortar)layer plays an elastic adjustment role,and its state performance directly affects the durability and maintenance workload for ballastless track.Ballastless track is long-term affected by vehicle impact and environmental temperature,therefore,CA mortar inevitably occurs damage,such as breakage and disengagement.In serious cases,it will significantly increase the dynamic force between the vehicle and the track,which will affect the comfort and safety of the vehicle.At present,there is still a lack of rapid detection theory and method for the CA mortar disengagement of ballastless track.Therefore,the systematic study of the influence law of CA mortar disengagement on the dynamic characteristics of vehicle and track and the development of the corresponding CA mortar disengagement detection method and intelligent sensing algorithm are of important theoretical value and practical guiding significance for ensuring the safety of vehicle operation and timely formulating the CA mortar maintenance plan.In this dissertation,CA mortar disengagement of CRTS II type slab ballastless track was considered as the starting point and the vehicle-track-subgrade coupling model which can simulate the CA mortar disengagement of arbitrary length was established.Based this,the influences of different CA mortar disengagement conditions on the dynamic characteristics of vehicle and track was studied and the intelligent sensing algorithm of ballastless track CA mortar disengagement was proposed.It can realize the CA mortar state inspection based on operating vehicles.The dissertation includes the following main contents:Aiming at the dynamic response simulation problem of vehicle-ballastless track under the combined action of CA mortar disengagement and track irregularity,this dissertation proposed to discretize CA mortar to make its distributed force on the track slab be transformed into concentrated fore,and established a vehicle-track coupling dynamic model considering the CA mortar disengagement action,which solved the practical problems that the existing dynamic model can not simulate the CA mortar disengagement in any condition and finite element model method with low simulation efficiency,limited line length and shorter wavelength of track irregularity.The reliable dynamic responses of the vehicle and track were obtained by employing the proposed model according to the new prediction-correction method based on Newmark,and the effects of different lateral disengagement degrees,longitudinal disengagement lengths,disengagement positions,and disengagement combinations and other conditions on the dynamic responses of vehicle and track were analyzed systematically.The sensitive conditions affecting the disengagement characteristics of CA mortar were obtained,and the limit value of the longitudinal disengagement length was given,which provides theoretical foundation and model basis for the research of CA mortar disengagement intelligent sensing algorithm based on vehicle dynamic responses.Aiming at the problem that the existing detection methods can not realize the rapid detection of CA mortar disengagement,this dissertation proposed a vehicular recognition algorithm of CA mortar disengagement based on time-frequency mixing feature extraction by transforming the CA mortar disengagement detection into pattern recognition problem.The algorithm takes the time domain combination features of the vehicle wheelset acceleration signal and the low-frequency features of power spectral density as the characteristic parameters and combines PSO-SVM to recognize and classify the CA mortar disengagement.In order to verify the robustness of the recognition algorithm,the simulation studies were carried out under the conditions of different vehicle running speeds,track irregularities,and signal-to-noise ratios.The results show that the proposed algorithm can recognize the CA mortar disengagement condition that the lateral direction is completely disengagement and can classify the different longitudinal disengagement lengths.Aiming at the difficulty in estimating the CA mortar lateral disengagement degree,this dissertation proposed an estimation algorithm of CA mortar lateral disengagement degree based on time-frequency analysis and model updating fusion.Firstly,by constructing the damage index sensitive to the change of CA mortar state,the multi-scale average wavelet energy value is used as the characteristic parameter,which can quickly locate the position of the track slab of CA mortar disengagement.Based this,by transforming the solution of the CA mortar lateral disengagement degree into the model parameter estimation problem,the spatial characteristics of vehicle dynamic responses and the prior knowledge were used to optimize the PSO iteration process to estimate the lateral disengagement degree of CA mortar,which improved the calculation efficiency and estimation accuracy.The simulation results of different vehicle running speeds,track irregularities and signal-to-noise ratios showed that the proposed algorithm can accurately estimate the lateral disengagement degree of CA mortar and has high robustness.Based on the vehicle-track coupling model,theoretical analysis and numerical simulation,the influences of CA mortar disengagement on the dynamic responses of vehicle and track and its intelligent sensing algorithm were systematically studied.The classification of CA mortar longitudinal disengagement length and the accurate estimation of lateral disengagement degree were realized respectively.This vehicle-mounted detection method provides a new idea for the development of CA mortar disengagement detection technology,and has important theoretical value and practical guiding significance.