中文摘要：

基于邻居搜索和模糊C均值方法,提出了一种新的鲁棒的聚类算法（NSFA）。该算法采用邻居搜索方法遍历相邻的数据,依据搜索距离确定聚类数目。NSFA方法引入内核函数,以提高其对噪声和异常点的鲁棒性。为了验证算法的鲁棒性,实验采用读取加速度计的数据,对比实验结果表明所提出的算法明显提高数据集的聚类性能,清晰划分加速度的变化趋势。

英文摘要：

With the aid of neighbor search and fuzzy c-means,a new robust clustering algorithm NSFA is presented.The proposed method is able to traverse the entire neighbordata,in terms of the search distance to determine the number of clustering. NSFA method adopts kernel functionto enhance its robustness to noises. In order to verify the robustness of the proposed algorithm,the experimental data are achieved from the accelerometer datasets. The simulation results show that NSFA algorithm obviously improves the performance of data clustering and the movement trajectory more distinctly.