Abstract: With the rapid development of information technology, data are becoming high-dimensional. How to accurately and efficiently cluster and properly apply these data is particularly important. Although the traditional Fuzzy C-Means clustering algorithm has a good clustering effect, the algorithm still fails to overcome the initialization sensitivity, besides when facing the problem massive high-dimensional network data, that algorithm is easy to fall into local extremum. In order to solve the problem, an adaptive fuzzy algorithm for fuzzy clustering anomaly detection is proposed, and applies it to the anomaly detection. The algorithm adds the concepts of distributed entropy and average bit distance in the classic bat algorithm, which makes the convergence speed of the algorithm greatly improved and can prevent the algorithm falling into local optimal solution. Simulation analysis shows that the algorithm is more stable in clustering and achieves promising detection performance.

Key words: Fuzzy C-Means（FCM）, adaptive bat algorithm, algorithm optimization, fuzzy clustering, anomaly detection

摘要： 随着信息网络技术的飞速发展，如何对规模庞大的网络数据准确高效聚类并合理应用显得尤为重要。虽然模糊C均值聚类算法（FCM）已具有良好的聚类效果，但其对初始化敏感，在处理高维大规模网络数据时易陷入局部极值问题还未被完全克服。为了解决这两个问题，提出一种分布熵和平均位距改进的自适应蝙蝠算法，利用该算法对模糊C均值的参数进行优化。在此之上，将自适应蝙蝠算法优化的模糊聚类应用于异常检测领域，提出了一种自适应蝙蝠算法优化的模糊聚类异常检测算法。理论分析和仿真实验表明，与前沿的粒子群优化FCM异常检测算法和FCM异常检测算法相比，该算法具有更好的聚类效果和检测性能。

关键词: 模糊C均值, 自适应蝙蝠算法, 算法优化, 模糊聚类, 异常检测