摘要
为准确掌握城市交通拥塞源内在的主要参数及扩散传播规律,以实现交通拥塞源科学管控.首先,引入用于大气污染物扩散的高斯烟羽模型并进行改进,将城市交通拥塞源划分为连续交通流和一系列间断性交通流,实现高斯烟羽模型结构解析;其次,利用Griewank、Schaffer和Rastrigin 3种测试函数对“单点源”参数反演算法进行测试,最终选用人群搜索算法;最后,通过交通拥塞源观测数据,从3个维度评估5种典型目标函数在不同参数数量(单、两、三)下的性能差异.研究结果表明:在单参数情形下,基于单位面积交通密度偏差平方和目标函数稳定性较好;反演源强相对偏差绝对值置信区间为38.38%±9.94%,小于50.00%实验次数占全部实验次数的84.52%,各目标函数稳定性均较好;在两参数反演源强情形下,基于对数变换单位面积交通密度均方根误差目标函数准确性最高,反演源强相对偏差绝对值置信区间为51.42%±9.84%,小于50.00%实验次数占全部实验次数的92.16%,在反演位置方面,基于单位面积交通密度偏差平方和目标函数准确性最好(反演位置偏差的绝对值为37.22 m±10.64 m),基于相关系数的目标函数稳定性最强(变异系数为0.022);三参数情形下,准确性反演结果和两参数较一致,除对数变换目标函数外各目标函数源强稳定性均较差,但位置稳定性均较好.
In order to accurately grasp the main parameters and diffusion and propagation laws inherent in urban traffic congestion sources and realize scientific control of traffic congestion sources,the Gaussian plume model for air pollutant dispersion was introduced and improved.The urban traffic congestion sources were divided into continuous traffic flow and a series of intermittent traffic flow,so as to realize the structural analysis of the Gaussian plume model.Then,three test functions,namely Griewank,Schaffer,and Rastrigin were used to test the“single point source”parameter inversion algorithm,and the seeker optimization algorithm was selected.Finally,the performance of the five typical objective functions with different numbers of parameters(one,two,and three)was evaluated in three dimensions based on the observed traffic congestion source data.The results show that in the one-parameter case,the stability of the objective function based on the sum of squared deviations of traffic density per unit area is better,and the confidence interval of the absolute value of the relative deviation of the inverse source strength is 38.38%±9.94%;the number of experiments less than 50.00% accounts for 84.52% of all experiments,and the stability of each objective function is better.In the two-parameter case of source strength inversion,the accuracy of the objective function based on the root-mean-square error of traffic density per unit area in the form of logarithmic transformation is the highest,and the confidence interval of the absolute value of the relative deviation of source strength inversion is 51.42%±9.84%;the number of experiments less than 50.00%accounts for 92.16%of all experiments.In terms of inversion location,the accuracy of the objective function based on the sum of squared deviations of traffic density per unit area is the best(absolute value of position deviation inversion is 37.22 m±10.64 m),and the stability of the objective function based on correlation coefficient is the strongest(coefficient of variation is 0.022).In the three-parameter case,the accuracy inversion results are more consistent with those in the two-parameter case,and the source strength stability of each objective function is poor except for the objective function in the form of logarithmic transformation,but the position stability is better.
作者
赵雪亭
胡立伟
ZHAO Xueting;HU Liwei(School of Transportation Engineering,Kunming University of Science and Technology,Kunming 650500,China)
出处
《西南交通大学学报》
北大核心
2025年第1期243-252,共10页
Journal of Southwest Jiaotong University
基金
国家自然科学基金项目(42277476,61863019)
云南省基础研究计划重点项目(202401AS070065)。
关键词
交通工程
高斯烟羽模型
启发式算法
参数反演
性能对比
traffic engineering
Gaussian plume model
heuristic algorithm
parameter inversion
performance comparison
作者简介
第一作者:赵雪亭(1997-),男,博士研究生,研究方向为城市交通管理与控制,E-mail:2287937331@qq.com;通信作者:胡立伟(1977-),男,教授,博士,研究方向为城市交通管理与控制,E-mail:liweihukm@kust.edu.cn。
