论文部分内容阅读
针对单一模糊C-均值聚类算法对初始聚类中心初值敏感性问题,引入同伦理论,提出了同伦模糊C-均值聚类算法。以三峡库岸研究程度较高的36个边坡为对象,采用同伦模糊C-均值聚类算法对边坡的稳定性进行分类,研究边坡最佳分类级数和算法的收敛性、可靠性。边坡聚类结果研究表明,同伦模糊C-均值聚类算法对初始聚类中心的选取没有明显的依赖性,是一个具有全局最优解的聚类方法,其结果明显好于单一模糊C-均值聚类算法。
Aimed at the initial sensitivity of initial clustering center to single fuzzy C-means clustering algorithm, the homotopy theory is introduced and the homotopy fuzzy C-means clustering algorithm is proposed. Taking the 36 slopes with high research level in the Three Gorges Reservoir as an example, the stability of the slope is classified by using homotopy fuzzy C-means clustering algorithm to study the convergence and reliability of the optimal classification level of the slope and the algorithm Sex. Slope clustering results show that the homotopy fuzzy C-means clustering algorithm has no obvious dependency on the initial clustering center selection. It is a clustering method with global optimal solution, and the result is obviously better than the single fuzzy C - Mean clustering algorithm.