基于遗传算法和BP神经网络的区域性公路交通量预测研究

发布时间：2018-11-12 12:05

【摘要】：公路交通是众多现代化交通运输中发展最迅速的运输方式,作为综合交通运输体系的主要组成成分,其具有基础性的地位,它是推动运输体系不断完善的主导力量。近年来,我国公路事业蓬勃发展,如何能够快速准确的进行交通量的预测是我们必须面对和解决的问题。预测模型的选择直接影响到了我们所需的数据资料以及预测的精度,本文根据对交通量预测方法和模型的研究,主要做了以下工作：首先本文分析了公路交通量预测在公路发展中的重要性,总结了公路交通量预测的发展趋势,并分析了常用的各种方法的优缺点。探讨了影响客运量和货运量的影响因素,采用相关系数法最终确定与客货运量相关的参数,分别对客运量和货运量进行预测。其次对BP神经网络和遗传算法进行了相关的分析和总结,指出了BP神经网络的缺陷,提出将遗传算法与BP神经网络算法相结合,即GA-BP模型。用遗传算法优化BP神经网络的权值和阀值,通过MATLAB建立模型实施模拟与预测,得出客货运量的预测值,并与实际值作比较,证明了预测方法的可行性。第三,根据预测的客货运量值,转换成标准车辆数,再采用合适的交通量分配方法,将其分配到相应路线上,与已有的调查点数据作比较,证明了该种模型用于交通量的预测的可行性。最后,对采用GA-BP网络模型预测交通量的局限性做了总结和说明,并也同时提出了相应的问题,为以后继续深入探讨提供思考。
[Abstract]:Highway traffic is the most rapidly developing transportation mode in many modern transportation. As the main component of the comprehensive transportation system, it has the basic position, and it is the leading force to promote the continuous improvement of the transportation system. In recent years, the highway industry of our country is booming, how to forecast the traffic volume quickly and accurately is the problem that we must face and solve. The choice of prediction model has a direct impact on the data we need and the accuracy of prediction. The main work is as follows: firstly, this paper analyzes the importance of highway traffic volume prediction in highway development, summarizes the development trend of highway traffic volume prediction, and analyzes the advantages and disadvantages of common methods. The influence factors of passenger and freight volume are discussed. The correlation coefficient method is used to determine the parameters related to passenger and freight volume, and the passenger volume and freight volume are forecasted respectively. Secondly, the BP neural network and genetic algorithm are analyzed and summarized, the defects of BP neural network are pointed out, and the combination of genetic algorithm and BP neural network algorithm, that is, GA-BP model, is put forward. The weight and threshold value of BP neural network are optimized by genetic algorithm. The prediction value of passenger and freight volume is obtained by using MATLAB model. The feasibility of the prediction method is proved by comparing with the actual value. Thirdly, according to the predicted passenger and cargo volume, the number of vehicles is converted to standard, and then the appropriate method of traffic flow distribution is used to distribute it to the corresponding route, and compare it with the existing survey data. It is proved that this model is feasible for traffic volume prediction. Finally, the limitations of using GA-BP network model to predict traffic volume are summarized and explained, and the corresponding problems are also put forward, which will provide some thoughts for further discussion in the future.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491.14

【参考文献】