基于CS-SNN的公交到站时间预测模型  

Prediction model of bus arrival time based on CS-SNN

作　　者：邝先验 罗会超 钟蕊 欧阳鹏 KUANG Xianyan;LUO Huichao;ZHONG Rui;OUYANG Peng(School of Electrical Engineering and Automation,Jiangxi University of Science and Technology,Ganzhou 341000,China)

机构地区：[1]江西理工大学电气工程与自动化学院,江西赣州341000

出　　处：《传感器与微系统》2021年第3期30-33,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(51268017,61463020);江西省教育厅科技项目(GJJ160609)。

摘　　要：针对公交到站时间预测不准以及工作日和周末运行特性差异的问题,提出了一种基于布谷鸟搜索算法的Spiking神经网络(CS-SNN)预测模型。通过分析和验证Spiking神经网络的性能特征,再采用寻优性能更强的CS算法优化Spiking神经网络的初始参数,最后利用行车历史数据对神经网络进行训练和建模来实现到站时间的准确预测,将该优化算法与未优化的SNN算法和Elman神经网络算法用MATLAB分别仿真测试。对比结果显示:无论工作日还是周末,CS-SNN预测模型对公交行程时间的预测均具有更高的准确性且结果更加稳定。A prediction model for spiking neural network based on cuckoo search algorithm(CS-SNN)is proposed to solve the problem of inaccurate bus arrival time and the difference of working characteristics between working days and weekends.By analyzing and verifying the performance characteristics of the spiking neural network(SNN),the cuckoo algorithm with better optimization performance is used to optimize the initial parameters of the Spiking neural network.Finally,the neural network is trained and modeled by driving history data to achieve accurate prediction of arrival time.It is predicted that the optimization algorithm is compared with the unoptimized SNN algorithm and the Elman neural network algorithm by MATLAB simulation.The comparison results show that the CS-SNN prediction model has higher accuracy for the bus travel time prediction regardless of the working day or weekend,and the results are more stable.

关 键 词：智能交通 公交到站时间预测 布谷鸟搜索算法 SPIKING神经网络 公共交通 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] U491[自动化与计算机技术—计算机科学与技术]