Intelligent Recognition Method Of Subgrade Differential Settlement Based On Dynamic Characteristics Of Vehicle And Track

This is a comprehensive research design which combines the railway Engineering and mathematical expertise, the article mainly done the following work and innovation:The dynamic response of the vehicle, the wheel-rail forces and fastener force and rail, as well as the subgrade due to subgrade settlement is calculated through the finite element model establishment by ABAQUS, the results show the inherent law between the dynamic responses, train speed and subgrade settlement which varies with the chord length and the amplitude; while the vehicle vertical acceleration is most sensitive to the differential settlement of subgrade. What’s more, the analysis also tells us where the maximum dynamic response in the settlement area is.Three types of smart identification neural networks were established based on GRNN network, BP network and BRF network, combined with the data calculated above. And the prediction performance the networks were discussed at the same time. The network use train’s vibration acceleration as input to predict the chord length and amplitude of subgrade settlement, through comparing, the RBF neural networks is better than the GRNN network and BP network in approach and forecast performance with a error of less than99.5%.On the basis of the above two studies, a database between differential settlement of subgrade and vehicle acceleration is proposed to be established, combined with the proposed neural network prediction method, a intelligent recognition system can be developed, and applied to Tim multiply instrument system, and through the inspection of vertical acceleration on the vehicle, the rail subgrade settlement can be known. Thus, the detection method of high-speed railway subgrade settlement can be optimized.