Multi-object Traffic Organization And Optimization Modeling And Simulating Based On AOP At Signalized Intersection

Supported by national 973 program of traffic congestion basic research program in big cities and Chang'an University Youth Science Fund of multi-object model for urban traffic organization and optimization, focusing on signalized intersection, on the question that moving stop line ahead led to different results in different areas, the paper made some related research from the aspect of signalized intersection traffic organization and optimization. Here are two goals. One of them is to make multi-object optimization model and solve it to get a idea combination of stop line, signal cycle and signal distributed by means of making dynamic and static traffic organization and optimization method match each other; the other is to get the best plan after comparing the result of various schemes visually and quantitatively in agent-oriented programming. The main achievements are follow:(1) built a multi-object optimization model and solved it by joining the classical mathematic ideal point method, PSO(Particle Swarm Optimization) and exhaustive algorithm in computer together. In order to avoid the question of ignoring smaller data in computing, the paper improved the evaluation function to balance each sub-object.(2) Based on the integrated idea, the paper put forth a population simulation framework of signalized intersection traffic organization and optimization. According to the theories of automaton in the compiler principle, this paper presented the designing and programming algorithms of the interface of static simulation models in urban traffic organization optimization. Data exchange was through files. As an example of DXF source file, it established a automaton of the line entity, tracing red information through State Transition Graph. At the same time, it also gave the class diagrams in UML(Unified Modeling Language) of static simulation models when they regenerate in target environment.(3) Set up related simulation models as follow: simulation environment model, signal model, car reaching model, car agent model, turn-left car trace model, and so on. Especially, it made a deep research for turn-left car trace model. It presented two different methods according to traffic state. For current or past state, data approximation can be adopt to get turn-left car trace model because traffic data can be got easily. And for future state, without any way to gain traffic data actually, only theory analysis can be used. Line-Cornu spiral-Cornu spiral-Line combination was used in this paper. After that, it reasoned a mathematics model about the relationship between stop line position and boundary radius at the joint point of two Cornu spiral segments.(4) Choose MT as random numbers creator by comparing of current random numbers creator through graphic analysis and statistics theory including hybrid linear congruential method, multiplicative linear congruential method, delay Fibonacci method, Mersenne Twister(MT),KISS method, combined creator, and so on; and further, created the corresponding random variable for simulation models.(5) Based on agent-oriented programming, signalized intersection traffic organization and optimization simulation system is developed, main UML class diagrams included. For a signalized intersection, we can use this system to choose the best plan after comparing the result of various schemes given visually and quantitatively. From two aspects of better approximation and easier programming, this paper studied the parameter estimation algorithms for simulation models. It presented maximum-likelihood method, normal number and B splines approximation and also made a comparison of the characteristics and conditions used.