基于FCM快速路交通状态判别加权指数研究

加权指数m是影响模糊C名)值聚类(fuzzy Cleans,FCM)的一个关键参数,为提高快速路交通状态模糊判別性能,针对m取值的问题提出了一种兼顾算法判別精度和聚类效果的优选方法。该方法以流量、速度为交通状态评价参数在不同加权指数m和样本量n下进行聚类分析,从算法判別精度、类内间距、类间间距、目标函数收敛性四个方面对m的最优取值进行了深入研究。以某市快速路为例別用MATLAB模糊逻辑工具箱分析实验数据的隶属度和聚类中心,以上四个方面在n×m种组合情形下综合分析,得出快速路交通状态模糊判別m的最优取值,并进一步验证了该方法的可行性。实验结果表明,以流量、速度为状态评价参数的快速路交通状态模糊判別加权指数m的最佳取值为2.25。

A Study of Weighting Exponent in Expressway Traffic State Estimation Based on Fuzzy C-Means

[Abstract] Weighting exponent m is an important parameter in fuzzy c-means (FCM) algorithm,for improving the performance of fuzzy expressway traffic state estimation, an method for accuracy of algorithm and clustering effect was proposed to optimal choice of m. In this method, flow and velocity were selected as traffic state evaluation parameters, cluster analysis was operated under different weighting exponent m and sample size n, and then an further study on optimal choice of m was made from algorithm accuracy, distance of simples to corresponding clustering centers , class distance, convergence of the objective function four aspects . Taken urban expressway as an example, MATLAB fuzzy logic of toolboxes was used to analysis membership and clustering center both of test data,a comprehensive analysis of the above four aspects was made under n by m kinds of combination cases, after that optimal choice of m was determined, and then it was verified to estimate the feasibility of the method. Experimental results show that the flow and velocity as evaluation parameters of expressway traffic state fuzzy estimation, the optimal choice of weighting exponent m is 2.25.