基于变异粒子群算法的公交线网分层优化研究

发布时间：2018-04-09 10:29

本文选题：城市公交线网　切入点：变异PSO算法　出处：《兰州交通大学》2014年硕士论文

【摘要】：随着经济和社会的发展，在我国，城市公共交通的发展相对缓慢，交通堵塞越来越严重，给居民出行带来不便，尤其是常规公交存在着线路布设的不合理，线路分布的不均衡、比较零散等问题，，不但降低了常规公交线网的运营效率，而且严重影响了公交线网的服务水平。目前大多数城市采用不分层法实现常规公交线网的优化布设，由于布设的线路分工不明确，衔接性差，缺乏整体性，甚至存在较多的交通盲区，因此，无法改善常规公交线网布设不合理的现状。基于以上公交线网优化所面临的问题，采用分层法对公交线网进行优化有着现实和重要的意义。本文主要将改进后的PSO(Particle Swarm Optimization,粒子群优化)算法与灰色预测法相结合，实现OD(Orgin-Destination,起点到终点)出行分布量的预测，在对公交线网分层优越性分析的基础上，建立公交线网的分层模型，并采用改进后的PSO算法对各层求解，形成合理的公交线网。首先，对连续和离散PSO算法进行改进。即针对PSO算法极易陷入局部最优的缺点，基于动态指数改进策略和遗传变异的思想，从改变算法的惯性权重和加入变异算子两个方面结合将算法改进为变异PSO算法，并通过具体的测试函数，分析变异PSO算法与传统改进的PSO算法的收敛性能。通过测试证明变异PSO算法具有良好的收敛性。其次，对公交线网采用分层法优化的优越性进行分析，并以兰州市的公交线网为研究对象，将变异PSO算法与灰色预测法相结合建立一种灰色变异粒子群组合预测模型，运用此模型在MATLAB软件中实现OD出行分布量的预测，通过与传统的增长计数法和重力模型法预测的结果进行比较，证明本文所建立的组合型预测模型的高精度性和适用性。最后，本文采用随机用户平衡法实现OD客流量分配，在此基础上实现主干线、次干线优化模型的建立及各层优化约束条件的确定；采用变异PSO算法对各层进行求解，完成主干线、次干线的优化布设；通过对优化后线网的重要指标计算分析，再次验证变异PSO算法具有良好的收敛性和分层法优化公交线网的优越性。运用分层优化法优化线网不仅能建立合理的公交线网，而且能提高公交线网的运营效率和服务水平。
[Abstract]:Compared with other problems, it not only reduces the operation efficiency of conventional bus network, but also seriously affects the service level of bus network.At present, most cities adopt the method of non-stratification to optimize the layout of conventional public transport network. Because of the unclear division of labor, poor cohesion, lack of integrity and even more traffic blind areas, the layout of the lines is not clear.Can not improve the normal bus network layout unreasonable status quo.Based on the problems faced by the above bus network optimization, it is of practical and important significance to adopt the hierarchical method to optimize the bus network.Firstly, the continuous and discrete PSO algorithms are improved.Aiming at the disadvantage that PSO algorithm is easy to fall into local optimum, based on the idea of dynamic exponent improvement strategy and genetic mutation, the algorithm is improved to mutation PSO algorithm by changing the inertia weight of the algorithm and adding mutation operator.The convergence performance of the mutated PSO algorithm and the traditional improved PSO algorithm is analyzed by testing function.The test results show that the mutation PSO algorithm has good convergence.Secondly, this paper analyzes the advantages of the hierarchical optimization of the bus network, and takes the public transportation network in Lanzhou as the research object, and combines the mutation PSO algorithm with the grey prediction method to establish a kind of grey variation particle swarm combination prediction model.This model is used to predict OD travel distribution in MATLAB software. By comparing with the results of traditional growth counting method and gravity model method, the high accuracy and applicability of the combined forecasting model are proved.Finally, this paper uses random user balance method to realize OD passenger flow distribution, on the basis of which the main trunk line, secondary trunk line optimization model and optimization constraints of each layer are established, and the variant PSO algorithm is used to solve each layer.Through the calculation and analysis of the important indexes of the optimized line network, it is verified that the mutated PSO algorithm has good convergence and the superiority of the hierarchical method to optimize the bus network.The hierarchical optimization method can not only establish a reasonable bus network, but also improve the operation efficiency and service level of the bus network.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U491.17

【参考文献】