In order to make full use of the actual expressway traffic congested information and to cluster the internal laws and characteristic changes of the traffic congestion more reasonably, an adaptive center and K-Means value (ACK-Means) clustering algorithm to cluster congested segments of expressways was proposed. ACK-Means algorithm assigns outliers reasonably by means of cluster density, cluster spacing and cluster strength, and considering the contingency of data samples. The ACK-Means algorithm determine the initial cluster center C and the number of clusters K-value adaptively. The data set was constructed based on the actual expressway congested information, and the ACK-Means algorithm was implemented by Python programming to cluster the expressway congested segments. The problem of clustering number K and initial clustering center C setting for the expressway congested segments was solved skillfully. Clustering results show that the ACK-Means algorithm realizes the unsupervised clustering of expressway congested segments, and the clustering results are based on the actual expressway congested information, which has higher practicability.