摘要
为提高城市路网动态交通规划准确度,提出一种基于多源传感器数据融合的城市路网动态交通规划方法。首先,采用固定、移动传感器获取浮动车数据、路网结构和互联网路况等多源数据;然后采用基于LSTM的速度数据融合方法进行多源传感数据融合;以多源数据融合结果为输入,采用PSO-RF短时交通状况预测模型同时预测多个路段交通状况,有效提取多路段间相互分流的特征信息;最后通过基于空间域分解的城市路网动态交通分配模型(DUE-DTA)对城市路网动态交通进行准确快速规划。实验证明,DUE-DTA模型规划路径配流满足DUE准则,其可取得较好的收敛性能,在收敛次数为28时,RG值仅为0.0027。交通流分配任务中,模型的拟合优度(R2)为0.9142,拟合精度较高,证明本模型可实现城市路网交通动态准确规划,满足公共交通管控的实际应用需求。
To improve the accuracy of urban road network dynamic traffic planning,a method for urban road network dynamic traffic planning based on multi-source sensor data fusion is proposed.First,fixed and mobile sensors are used to obtain multi-source data such as floating car data,road network structure and Internet road conditions;Then,a speed data fusion method based on LSTM is used for multi-source sensor data fusion;Using multi-source data fusion results as input,the PSO-RF short-term traffic condition prediction model is adopted to simultaneously predict the traffic conditions of multiple road sections,effectively extracting the feature information of mutual diversion between multiple road sections;Finally,the urban road network dynamic traffic allocation model based on spatial domain decomposition(DUE-DTA)is used to accurately and quickly plan the dynamic traffic of the urban road network.Experimental results have shown that the DUE-DTA model satisfies the DUE criterion for path allocation planning and can achieve good convergence performance.At a convergence frequency of 28,the RG value is only 0.0027.In the task of traffic flow allocation,the goodness of fit(R2)of the model is 0.9142,indicating high fitting accuracy.This proves that the model can achieve dynamic and accurate planning of urban road network traffic,meeting the practical application needs of urban and rural public transportation control.
作者
毛盈盈
MAO Yingying(Southeast University Chengxian College,Nanjing 210088,China)
出处
《自动化与仪器仪表》
2026年第1期91-95,共5页
Automation & Instrumentation
基金
2021年江苏省高等学校自然科学研究面上项目《公共服务均等化视角下城乡公共交通可达性研究》(21KJD580004)
2020年东南大学成贤学院“青年教师科研发展基金”项目《公共服务均等化视角下城市公共交通可达性评价》(z0031)
2023年东南大学成贤学院“国家级科研项目培育基金”项目《双碳背景下基于多源数据的步行交通设施配置方法研究》(2023NCF001)。
关键词
多源传感器
数据融合
城市路网
交通规划
交通状况预测
multi source sensors
data fusion
urban road network
transportation planning
traffic condition prediction
