基于MFCM的改进聚类算法及其在交通中的应用

在交通流状态模糊化的过程中,对已有的交通模糊控制研究引入了太多的主观因素.为了解决这个问题,提出了一种基于MFCM算法的分级递减聚类算法,利用MFCM算法寻找类中心,再自适应确定该类中心的隶属度阈值,将聚类进行分级处理,实现未知类别数数据集的聚类.将改进算法应用到交通流状态聚类中,可以更科学地确定交通流状态的聚类数和各类模糊隶属度函数的结构等,最后,通过算例,说明了该算法对于未知聚类数及服从高斯分布的数据集具有聚类效果好、收敛速度快的特点.

Improved clustering algorithm based on modified fuzzy C-means applied to the traffic

There were many subjective factors in the fuzzification of traffic flow states of existed research about traffic fuzzy control.A hierarchical subtractive clustering algorithm based on modified fuzzy C-means is proposed.It finds that the cluster centers using modified fuzzy C-means clustering algorithm give the membership threshold of the centers adaptively,perform the clustering progressively and then the clustering for the data set with unknown number of clusters is completed.The improved algorithm is applied to the fuzzification of traffic flow states.It can give the number of categories of traffic flow states and the configuration of all kinds of membership functions more accurately.An example is given to show that the proposed algorithm clusters the Gaussian distributed data patterns without a prior information about the number of clusters with good effectiveness and fast clustering convergence.