结合车位预测的智能停车系统的研究

发布时间：2018-08-27 09:23

【摘要】：近年来,机动车保有量的迅猛增长使得停车难的问题日益凸显。造成停车难的原因是多方面的,主要表现在停车场车位信息闭塞、停车场人工收费效率低下以及停车场车位流转率不高等。为了减轻停车难的压力,引入采用先进互联网技术以及智能设备的智能停车系统升级改造传统停车系统是大势所趋。传统停车系统无法提供实时车位信息,不支持在线缴纳停车费用,也没有车位预测功能。针对这些问题,在带有车牌识别系统的线下收费软件的基础上,本课题基于安卓系统开发了一款具有车位预测功能的智能停车系统。最终实现的系统支持车位信息实时发布、用户在线缴纳停车费用以及车位预订等功能。在开发过程中,首先,结合前期对用户以及停车场的调查结果,对系统进行需求分析并讨论了系统实现所涉及的相关框架与技术。系统通信协议为HTTP协议,数据交互格式选用JSON。然后,根据具体需求,系统被划分为我要停车、停车助手以及个人中心等功能模块。进一步地,针对各个功能模块设计对应的数据表,按模型-视图-控制器模式完成系统的设计与实现。同时,为了提高停车场车位的流转率以及车位预测的准确性,分别基于灰色模型和神经网络设计了空闲车位的预测方法,并根据误差变化动态调整学习速率以优化神经网络。为了验证预测方法的有效性,针对成都某停车场的空闲车位情况做了预测仿真,并就仿真结果进行了分析讨论。在此基础上,系统最终选用学习速率动态调整的神经网络作为空闲车位预测方法。最后,对该智能停车系统进行功能测试、非功能测试以及用户满意度调查。结果表明该系统具有较好的实时性、稳健性、兼容性以及较高的用户满意度。采用该系统之后,停车场车位信息更加透明化、车位流转率更高、缴费方式更加多样化、寻找车位所花时间更加少。间接地改善了交通拥堵状况,并在一定程度上减轻了环境污染。
[Abstract]:In recent years, with the rapid growth of vehicle ownership, the problem of parking difficulties has become increasingly prominent. There are many reasons for parking difficulties, such as the block of parking space information, the low efficiency of parking lot manual charge and the low circulation rate of parking space. In order to reduce the pressure of parking difficulties, it is the general trend to introduce the intelligent parking system which adopts advanced Internet technology and intelligent equipment to upgrade and transform the traditional parking system. Traditional parking systems can not provide real-time parking information, do not support online parking fees, and do not have parking prediction function. In order to solve these problems, an intelligent parking system with the function of parking space prediction is developed on the basis of offline charging software with license plate recognition system. The system supports real-time distribution of parking space information, online parking fees and parking reservation. In the process of development, first of all, combined with the previous survey results of users and parking lots, the requirements of the system are analyzed and the related framework and technology involved in the implementation of the system are discussed. The communication protocol of the system is HTTP protocol, and the format of data exchange is JSON.. Then, according to the specific needs, the system is divided into my parking, parking assistant and personal center and other functional modules. Furthermore, the corresponding data table is designed for each functional module, and the system is designed and implemented according to the model-view-controller mode. At the same time, in order to improve the flow rate of parking space and the accuracy of parking space prediction, the prediction method of free parking space is designed based on grey model and neural network respectively, and the learning rate is dynamically adjusted according to the error change to optimize the neural network. In order to verify the effectiveness of the prediction method, the free parking space in a parking lot in Chengdu is predicted and simulated, and the simulation results are analyzed and discussed. On this basis, the system finally chooses the learning rate dynamic adjustment neural network as the free parking prediction method. Finally, the function test, non-function test and user satisfaction survey of the intelligent parking system are carried out. The results show that the system has good real-time, robustness, compatibility and high user satisfaction. With this system, the parking space information is more transparent, the parking space flow rate is higher, the payment method is more diversified, and the time to find the parking space is less. Indirectly improve the traffic congestion, and to some extent reduce environmental pollution.
【学位授予单位】：四川师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495;U491.7

【参考文献】