VISSIM交通仿真模型参数校正技术研究

发布时间：2018-04-25 23:23

本文选题：微观交通仿真 + 参数校正　；参考：《吉林大学》2015年硕士论文

【摘要】：随着城市的快速发展，居民私家车保有量逐年递增，交通拥堵问题已经在很多城市引起了有关部门的极大关注。微观交通仿真软件是评估交通解决方案的有力工具，因此软件模型的精确性就显得至关重要。目前，我国大多数研究机构所使用的交通仿真软件都从国外引进，因此依据我国实际交通运转状况对软件模型参数进行校正是开展其余工作的前提和基础。以往的参数校正算法大多采用遗传算法，但是遗传算法在迭代过程中会耗费大量的时间，同时目前大多数校正的方法都是独立的程序，并未达成参数校正的自动化，为了解决上述问题，本文建立了基于改进方法的VISSIM参数自动校正体系。在研究过程中，主要实现了以下几个方面的工作：首先，，经过对大量文献的阅读和整理，归纳出目前对于参数校正工作的研究正沿着两条主线展开，而本文的研究重点也放在对于模型校正算法的研究上。本文以VISSIM仿真软件为例，对软件的核心模型——跟驰模型和换道模型的重要参数进行了详细地说明，然后对参数校正过程中评价指标的选取和待校正参数的选取方法做了具体地介绍。其次，本文以遗传算法为参数校正方法，利用训练好的广义回归神经网络模型预测仿真软件VISSIM的输出结果，这样就避免了在遗传算法迭代过程中需要反复运转仿真软件造成的时间浪费。这部分也是论文的核心之一。之后本文以北京市中关村一街为实例，对上述参数校正方法进行了实例验证，结果证明，该方法能够有效提高参数校正的效率，并且符合对模型精度的要求。接着，文章建立了交通仿真软件自动校正体系，达成了参数校正的流程化和自动化，用户能够经过简单的图形界面达成对参数校正流程的控制，同时能够获得更加直观的校正前后参数和评价指标的对照状况。最终，文章对上述建立的自动校正体系进行了实例的验证，验证的区域是江苏省无锡市新区的主干道——菱湖大道从高浪路到震泽路路段，以平均行程时间为评价指标，以浮动车跟车和实地调查的方式对交通流数据进行了采集，并在VISSIM平台上建立了仿真模型。运转结果表明，用户能够经过该体系达成对模型参数校正的空子，并且模型校正的结果在可接受的范围内。
[Abstract]:With the rapid development of cities, the number of private cars is increasing year by year. Traffic congestion has aroused great concern of relevant departments in many cities. Microscopic traffic simulation software is a powerful tool for evaluating traffic solutions, so the accuracy of the software model is very important. At present, the traffic simulation software used by most research institutions in our country is imported from abroad, so it is the premise and foundation of the other work to correct the parameters of the software model according to the actual traffic operation in our country. In the past, most of the parameter correction algorithms used genetic algorithm, but the genetic algorithm in the iterative process will cost a lot of time, and most of the current correction methods are independent procedures, did not achieve the automation of parameter correction. In order to solve the above problems, an automatic correction system of VISSIM parameters based on the improved method is established. In the course of the research, the following aspects of the work are realized: First of all, through reading and sorting out a large number of documents, we conclude that the research of parameter correction is being carried out along two main lines, and the research emphasis of this paper is also on the research of model correction algorithm. Taking the VISSIM simulation software as an example, this paper gives a detailed description of the important parameters of the core model of the software, that is, the car-following model and the changing channel model. Then the selection of the evaluation index and the method of selecting the parameters to be corrected in the process of parameter correction are introduced in detail. Secondly, using the genetic algorithm as the parameter correction method, the trained generalized regression neural network model is used to predict the output of the simulation software VISSIM. In this way, the time waste caused by running simulation software repeatedly in the iterative process of genetic algorithm is avoided. This part is also one of the core of the paper. Then this paper takes Zhongguancun first Street in Beijing as an example to verify the above parameter correction method. The results show that the method can effectively improve the efficiency of parameter correction and meet the requirements of model accuracy. Then, the automatic correction system of traffic simulation software is established, and the process and automation of parameter correction are achieved. The user can control the process of parameter correction through a simple graphical interface. At the same time, the comparison of parameters and evaluation indexes before and after correction can be obtained more intuitively. Finally, the paper verifies the automatic correction system, which is the main road of Wuxi City, Jiangsu Province, from Gaolang Road to Zhenze Road, and takes the average travel time as the evaluation index. The traffic flow data were collected by floating vehicle following vehicle and field investigation, and the simulation model was established on VISSIM platform. The operation results show that the user can achieve the model parameter correction through the system, and the model correction results are acceptable.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;TP391.9

【参考文献】