论文部分内容阅读
为获取居民公交出行的换乘信息,设计了一套基于多分类支持向量机(Multi-class Support Vector Machine)的公交换乘识别方法。通过融合GPS数据和公交IC卡数据获取训练样本,利用多分类支持向量机进行样本训练,选取最佳训练样本量,并采用网格搜索法结合粒子优化算法对模型参数进行标定,以获取最优SVM分类模型。测试结果显示模型分类精度可达90%。以佛山市公交车GPS数据和IC卡数据对算法进行验证,并获取公交换乘量、公交换乘比例等基本换乘数据。结果表明:算法可在少样本条件下完成公交换乘识别,且分类识别精度高,尤其适用于公交线网复杂的大城市公交换乘识别,有助于在公交前期规划时进行线路布设和枢纽选址。
In order to obtain the transfer information of residents’ public transportation, a set of bus transfer recognition method based on Multi-class Support Vector Machine is designed. Through the integration of GPS data and bus IC card data to obtain training samples, the use of multi-class support vector machine for sample training, select the best training sample size, and grid search method combined with particle optimization algorithm to model parameters to obtain the optimal SVM classification model. Test results show that the model classification accuracy of up to 90%. The bus GPS data and IC card data of Foshan are used to verify the algorithm and obtain the basic transfer data such as bus transfer and bus transfer ratio. The results show that the algorithm can accomplish bus transfer recognition under few samples and has high classification accuracy, especially suitable for transit transfer recognition in metropolitan areas of public transit networks, which is helpful for route layout and hub planning in early public transport planning Site selection.