Wind-vehicle-bridge Coupled Vibration Analysis And Control Of Single-level Rail-cum-road Bridge

In recent years,the development of bridge engineering and the shortage of bridge site resources have promoted the application of rail-cum-road bridges.Due to its unique wind environment on the bridge deck and the prominent dynamic and aerodynamic interference between railway and highway,the single-level arrangement is different from the other ones.Hence,the research of this dissertation is carried out on the wind-resistance performance of traveling vehicles and traffic control of single-level rail-cum-road bridges.(1)Based on 1:20 large scale segmental model wind tunnel test,the aerodynamic performance of car-train-bridge system of a single-level rail-cum-road bridge under crosswind was studied.The aerodynamic forces of four kinds of road vehicle as well as CRH3 train at different positions on the bridge deck are tested.The aerodynamic variations caused by the windshield effect of upstream vehicles in the process of overtaking on the windward lane are discussed,and the influence of streaming around effect of upstream vehicles on the aerodynamic performance of downstream vehicles are researched.(2)With computational fluid dynamics numerical simulation and sliding mesh technology,the aerodynamic forces of van in the process of meeting the train and chased by the train are calculated.The numerical accuracy is verified by a moving model test.Through visualization technology,the variation feature of the flow filed is discussed,based on which the interaction mechanism between the van and train is analyzed.Furthermore,based on the formulas about the pressure wave produced by the travelling trains,the distance and the vehicle velocity,the relationship between the aerodynamic variation amplitude of the van and the transverse distance as well as the relationship between the aerodynamic variation amplitude of the van and the vehicle velocity,are explored.Finally,the proportion of train velocity term,van velocity term and coupled velocity term in aerodynamic variation amplitude is calculated,and the promotion and suppression effects of each velocity term on the aerodynamic variations are revealed.(3)To explore the wind-resistant performance of travelling vehicles on single-level railcum-road bridge,the aerodynamic forces of road vehicles under crosswind and train-induced wind are calculated by numerical model.Static model wind tunnel test is utilized to verify the accuracy on simulating the wind environment on the bridge deck,and moving vehicle model test is employed to verify the accuracy of the dynamic grid technology and the rationality of the mesh resolution around the bridge deck system.The effects of vehicle type and vehicle position on the superposition of crosswind and train-induced wind were studied.Finally,a quick method is proposed to obtain the aerodynamic forces of vehicles rendezvousing with train under arbitrary wind speed through several numerical cases.(4)A car-train-bridge coupled vibration system,which has considered the dynamic and aerodynamic interference between road vehicles and trains,was established.By comparing the results of present coupled vibration system with the ones of previous train-bridge coupled vibration software and the ones of previous car-bridge coupled vibration software,the calculation accuracy is validated.After analyzing the influence of local vibration produced by trains on the road vehicles,the reasonability of applying the beam-link element model on the vehicle-bridge coupled calculation is verified.The influence of dynamic and aerodynamic disruption of trains on different types and positions of road vehicles is discussed,and the relationship between vehicle response extremum and load variation curve is analyzed.Based on the vehicle response,the proposed method which could quickly obtain the aerodynamic load of road vehicle is adjusted.(5)An ultra-short time wind speed prediction method based on wavelet analysis,deep learning and ensemble learning is proposed.Taking the wind speed series in a complex mountainous area as test sample,the superiority of the proposed method was verified by comparing the calculation accuracy with the ones of other hybrid models.The accuracy of tree model is improved by high dimensional feature extraction method of deep belief network,and the superiority of tree model in processing high dimensional features compared with back propagation neural network of the top layer of deep belief network is verified.Boosting algorithm and Bagging algorithm of ensemble learning algorithm are compared in the aspect of accuracy and stability.