山地城市公交到站信息预测研究

发布时间：2018-04-29 06:36

本文选题：城市公共交通 + 预测　；参考：《重庆交通大学》2014年硕士论文

【摘要】：随着社会的不断进步，交通拥堵已经成为21世纪各大城市必须面临的重大难题，而解决城市交通拥堵的重要途径就是大力发展公共交通。公交车辆作为公共交通的基本组成部分在缓解城市交通拥堵，提高城市道路利用率上发挥了极大的功效。城市公交信息服务系统是实现城市智能公交系统的基础，它以“为出行者出行服务”为目的。因此，必须旅客的角度出发，发布旅客重点关心的出行信息。通过对乘客进行实际调查后得知，其中公交车辆到站信息是出行最为关心的信息之一。所以，集中优势力量，利用所获得的交通采集数据，分析和发布出行者最为关心的公交车到站信息是实现城市公共交通系统信息化的重要内容。论文首先对公交车辆定位信息的采集系统进行了分析，介绍了GPS定位的基本原理，对所采集到的GPS定位信息的原始数据所包含的内容以及每个参数所对应的实际意义进行了详细的说明。随后对公交定位信息产生误差的原因从多个角度进行了研究，并从其误差产生的根源出发，提出了相对应的补偿方案。接着，为了提高公交到站信息预测的精度，提出公交实时数据匹配所采用的方法，其中包括对公交线路路段的划分方法、公交路线的线性化处理方法以及通过公交定位数据中的方向角一项来对公交车辆行驶方向进行判别的方法。在此基础之上，对公交到站时间的影响因素进行了分析，，依据公交车辆的行驶特点，将其总的行驶时间分为了三个部分，分别是路段行驶时间、交叉口延误时间与到站停靠时间，并就每个部分的影响因子的重要程度和产生影响的原因就行了研究。针对各种影响因素的差异，提出利用BP神经网络模型来对公交到站时间进行预测，并对原有的神经网络数据训练算法进行优化，以提高公交到站时间预测的精度和降低模型训练所需要的时间。最后，利用重庆市601路公交车的实时定位数据进行实验，实验结果表明论文提出的优化模型样本训练速度有了较大的改善，预测精度也在可接受范围内，在实际应用时能有较好的发挥。
[Abstract]:With the continuous progress of society, traffic congestion has become a major problem that every major city must face in the 21st century, and the important way to solve urban traffic congestion is to develop public transportation. As a basic part of public transport, public transport vehicles play a great role in alleviating urban traffic congestion and improving the utilization ratio of urban roads. The urban public transport information service system is the foundation of realizing the urban intelligent public transport system, which aims at serving the travelers. Therefore, it is necessary to release the travel information which the passengers are concerned about from the point of view of the passengers. Through the actual investigation of passengers, we know that bus arrival information is one of the most concerned information. Therefore, it is an important content to realize the informatization of urban public transportation system by concentrating the advantages, using the traffic data collected, analyzing and publishing the bus arrival information that the travelers are most concerned about. Firstly, the paper analyzes the collection system of public transportation vehicle positioning information, and introduces the basic principle of GPS positioning. The contents of the original data of the collected GPS location information and the practical meaning of each parameter are explained in detail. Then, the causes of the errors of public transportation positioning information are studied from several angles, and the corresponding compensation scheme is put forward from the root of the errors. Then, in order to improve the accuracy of bus arrival information prediction, the method of bus real-time data matching is put forward, including the method of dividing the bus route. The linearization method of bus route and the method to distinguish the driving direction of public transport vehicle by the direction angle in the location data of public transport. On this basis, the influence factors of bus arrival time are analyzed. According to the driving characteristics of public transport vehicles, the total travel time is divided into three parts. The intersections delay time and arrival time, and the importance of each part of the impact factors and the causes of the impact are studied. According to the difference of various factors, the BP neural network model is proposed to predict the bus arrival time, and the original neural network data training algorithm is optimized. In order to improve the accuracy of bus arrival time prediction and reduce the time required for model training. Finally, using the real-time positioning data of the 601 bus in Chongqing, the experimental results show that the training speed of the optimized model has been greatly improved, and the prediction accuracy is within the acceptable range. In the actual application can have a better play.
【学位授予单位】：重庆交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495;U491.17

【参考文献】