In view of the traditional data stream clustering algorithm clustering information loss, inaccurate faults, a data stream clustering algorithm based on the dimension maximum entropy is proposed. Dynamic data in the sliding window are divided into different dimensions by using data histogram. The maximum entropy of different dimension is calculated to classify dimension spaces to form a cluster dimensions. Data are gathered into small clusters of the same dimension of cluster. By comparing the size of the cluster of information entropy and its distribution features, outlier detection of data stream is realized. This method improves the clustering speed, and overcomes the traditional shortcomings of the data stream clustering algorithm information loss. Experimental results show the effectiveness of the proposed algorithm.