Study On3D Spatial Intelligent Route Selection Approach For Railway Alignments

In the procedure of railway design, horizontal and vertical alignments are of vital importance, as it will directly affect the investment of construction and operation of a line. Therefore, study of profile optimization can bring practical social and economic benefits. Existing models for optimizing railway or highway alignments were reviewed and the advantages and disadvantages of different approaches were identified. An improved approach for solving three-dimensional alignment optimization problems which integrates dynamic programming with Powell method was developed. The approach presented in this paper mainly performs as following:(1) The methodology presented in this paper firstly formulated the optimization of alignment as a network problem by establishing a three-dimensional search grid in the study area. Then dynamic programming as one of the most well-developed network optimization techniques for solving shortest path problems was used to find the best corridor alignment within the vast study area. As dynamic programming can only obtain one single global optimum, which can not satisfy the scenario-diversity demand of railway alignment selection, a double-direction searching strategy in dynamic programming was proposed to obtain a group of scenarios ranking by their corresponding comprehensive cost.(2) In theory, dynamic programming can guarantee that the path obtained must be the global optimum, subject to the grid established. Dynamic programming is not perfect for continuous search space, as it can only obtain discrete solutions that confined to certain relatively coarse grid of points. Powell’s direction acceleration method which is desirable for multiple function optimizations was introduced to refine the preliminary alignment so that the resulting alignment could deviate from the grid points to reach the global optimum. Powell method which has a fast convergence speed is so far the most effective direct method; however, it is an effective unconstrained problem solving tool which is not suitable for dealing with constrained problems. In this paper constrained problem was converted to unconstrained problem by adding penalties to the objective function to avoid constraint violations. With the theory and method reviewed in this paper, a three dimensional spatial intelligent route selection system was developed. Examples were presented to illustrate the proposed model and the performance of the solution algorithm. Experiments demonstrated that this approach can successfully obtain from the static and certain circumstance optimal scenarios of both horizontal and vertical alignments without violating the corresponding constraints and scenarios obtained with this method could effectively adjust to the terrain and avoid barriers. Both horizontal and vertical alignments seemed reasonable and fit the topography quite well.