基于因子分析与改进K-means聚类的交通状态判别

Traffic state discrimination based on factor analysis and improved K-means clustering

为了提高城市快速路交通运行状态判别的精确度,提出一种基于因子分析与改进K-means聚类的城市快速路交通状态判别方法。首先运用因子分析对流量、速度、占有率、密度、行程时间、饱和度、占有率/流量、占有率/速度等8个交通流参数进行了相关性分析及适用性检验,提取了流量、速度和占有率/速度3个交通状态判别指标;其次通过建立交通状态综合评价函数对K-means聚类算法进行改进,将快速路交通状态聚簇为4类;最后使用快速路的实测数据完成算法的实例验证和对比分析。结果表明,基于因子分析与改进K-means聚类算法的快速路交通状态判别率为97.21%,误判率为0.74%,相对于传统K-means聚类算法判别精度提高了8.13%,误判率降低了1.05%。

In order to improve the accuracy of traffic state discrimination of urban expressway,a traffic state discrimination method based on factor analysis and improved K-means clustering was proposed.First,using factor analysis,the correlation analysis and applicability test were carried out on 8 traffic flow parameters of flow,speed,occupation,density,travel time,saturation,occupation/flow,occupation/speed.The three traffic state discrimination indexes of flow,speed and occupation/speed were extracted.Then,K-means clustering was improved by establishing the comprehensive evaluation function of traffic state,and the traffic state of expressway was clustered into 4 categories.Finally,the actual data of expressway was used to verify and compare the algorithm.The results indicate that the traffic state discrimination rate of this method is 97.21%and the error rate is 0.74%.The discriminant accuracy is 8.13%higher than the traditional K-means clustering,and the error rate is reduced by 1.05%.