Virtual Inversion Of Vibration Source On Surface Of Single Soil Layer Overlaid On Elastic Half-Space

Traffic congestion is now one of the most important problems in city development of our country. An effective method to solve this problem is developing urban track traffic. Thus, study on environmental vibration and its prevention has been an urgent problem.Inversion of vibration source from urban track traffic by observed data on an array is a quite new idea. The virtual inversion is adopted in this dissertation, in the process, a vibration source function is given, and vibration on each point in the array is calculated then. Based on the vibrations at the array, the parameters of the source function are inversed and are furthermore checked with the given ones to validate the feasibility of this idea.Firstly, site condition is simplified as a model with single soil layer overlaid on elastic homogeneous half-space. The analytical solution of vertical displacement field on the surface of the model from pulse loading on the surface is derived by means of transfer matrix and Laplace-Hankel combined transform. In the derivation all layers contacted with each other completely. Afterwards, the numerical realization of the solution is introduced in detail, five possible problems in the process are discussed and the method to manage them is presented, and a Fortran program is compiled. The numerical result by degenerating the model into elastic half-space is checked with classical solution to validate the reliability of the calculation in this dissertation.Then, the evolution and the state-of-the art of Genetic Algorithm are reviewed briefly, as well as the principles and existing problems in the basic Genetic Algorithm. The gist and the procedure in the Micro Genetic Algorithm are introduced in detail, the generic strategy for inversion is designed, the program for inversion is also compiled. The reliability of the program is verified by four test functions.Finally, displacement time histories at six points in a rectangular array are calculated from the given site model with single soil layer overlaid on elastic half-space, given array layout and given triangular pulse input. The information in the time histories are analyzed, frequency dispersion curve at the site is acquired and is checked with those derived directly from frequency dispersion equation. Micro Genetic Algorithm is adopted in the virtual inversion of source parameters. Target curves are constructed respectively by the time history at a point, the auto-power spectrum at a point, the cross-power spectrum between a pair of points in the array. The results are checked with the given source parameters, and show a good stability and efficacy. The information on two points can improve the inversion results. As a conclusion, the idea to inverse traffic vibration source from the observation data at an array is feasible, if the target function is constructed well.