提出一种基于免疫原理、对不确定数据流进行聚类的算法——IUMicro.IUMicro针对不确定数据流上元组级不确定性问题,引入动态更新以适应数据变化的免疫模型,其中包括一种有效的在线收集数据流统计信息的B细胞特征结构及其更新策略.为兼顾元组存在概率与元组间的距离两方面因素,定义概率识别半径,为每个不断到达的数据元组找到合理的候选簇.离线聚类根据免疫细胞识别区域的空间关系,进行任意形状的无监督聚类.实验结果表明,IUMicro能有效抑制噪声,具有良好的聚类质量和较快的处理速度. An algorithm for clustering uncertain data flows based on immune principles is proposed.iumicro.IUMicro Aiming at the uncertainty of tuples at uncertain data flow, an immune model that incorporates dynamic updates to adapt to data changes, including an An Effective B-cell Signature Structure for Collecting Data Flow Statistics Online and Its Updating Strategy To define the probability identification radius for both the existing probability of a tuple and the distance between tuples, we find for each arriving data tuple Reasonable candidate clusters.Offline clustering unsupervised clustering with arbitrary shape according to the spatial relationship of immune cell recognition regions.The experimental results show that IUMicro can effectively suppress the noise and has good clustering quality and fast processing speed.