基于遗传模拟退火算法的TRANSMODELER仿真模型参数标定研究

大多微观仿真软件是针对国外交通流特性而开发的交通仿真软件,而我国的交通特性与国外相比存在一定的差异,因此在运用仿真软件进行交通仿真之前,需要针对我国城市路网的实际交通调查结果,对相关参数进行标定.其目的是提高仿真模型的精度,使仿真结果更真实、更准确.文中设计了详细的参数标定流程,利用参数敏感性分析选取所需要标定的参数,再引入遗传模拟退火算法进行参数最优组合的搜索,并设计其功能界面,利用TRANSMODELER的二次开发功能,将参数标定的算法植入其仿真模块,实现仿真、评价的一体化和自动化.以昆明市一二一大街、学府路及相关联路道路为示范路网,验证所建立参数标定流程与算法的有效性与可行性.

Study on Parameter Calibration for Simulation Model TRANSMODELER Based on Genetic Simulated Annealing Algorithm

Most of simulation software were developed for the foreign traffic flow characteristics,but traffic characteristics of our country have some differences from abroad,so before making traffic simulation,it need to calibrated relevant parameters according to traffic survey results of our road network.The purpose is to improve the accuracy of the simulation model,and make the simulation results more realistic and more accurate.This paper conducted a detailed parameter calibration process,used the way of parameter sensitivity analysis to choose the parameters which need to calibrate,then introduced genetic simulated annealing algorithm to search optimal parameters combination.Then its function interface was designed,parameters combination algorithm was implanted to the simulation module by using the secondary development function of TRANSMODELER,to achieve integration and automation of simulation and evaluation.Finally,Xuefu Road-Yieryi street and some other related roads were taken as examples to verified the validity and reliability of the established parameter calibration process and parameter calibration algorithm.