海洋数据下的密度自适应聚类算法

Density adaptive clustering algorithm in ocean data

针对DBSCAN算法需要人工设定参数,且数在对不同疏密度的数据敏感度较低以及处理多维多密度的海洋数据时鲁棒性欠佳的问题,提出一种基于K-均值模型的多密度自适应聚类算法AM-DBSCAN(adaptive multi-density DBSCAN algorithm)。采用K-均值模型对数据进行初次聚类,分别以结果簇中距离最远两点的平均值及最小簇的样本数作为DBSCAN算法中的邻域半径(Eps)及邻域样本阈值(Minpts);以最短路径原则改进DBSCAN算法中Eps邻域判定方式,提高算法全局的可靠性及稳定性。实验结果表明,相对于DBSCAN聚类算法,AM-DBSCAN算法在处理密度不均的数据时在聚类准确度和聚类效率方面有所提升。

Aiming at the problems that the DBSCAN algorithm needs to set parameters manually,and the number is less sensitive to different density data and less robust to multi-dimensional multi-density ocean data,a multi-density based on K-means model was proposed.Adaptive clustering algorithm AM-DBSCAN(adaptive multi-density DBSCAN algorithm)was proposed.The K-means model was used to cluster the data for the first time.The average of the two far points in the result cluster and the number of samples of the smallest cluster were used as the neighborhood radius(Eps)and neighborhood samples threshold(Minpts)in the DBSCAN algorithm.The Eps neighborhood decision mode in the DBSCAN algorithm was improved using the shortest path principle,which improved the overall reliability and stability of the algorithm.Experimental results show that compared with the DBSCAN clustering algorithm,the AM-DBSCAN algorithm improves the clustering accuracy and clustering efficiency when dealing with uneven density data.