基于聚类云模型的小样本数据可信度评估  

Credibility Evaluation Method of Small Sample Data Based on Cluster Cloud Model

作　　者：王建敏[1] 吴云洁[2] Wang Jianmin;Wu Yunjie(Technology and Engineering Center for Space Utilization,Chinese Academy of Sciences,Beijing 100094,China;School of Automation Science and Electrical Engineering,Beijing University of Aeronautics and Astronautics,Beijing 100083,China)

机构地区：[1]中国科学院空间应用工程与技术中心,北京100094 [2]北京航空航天大学自动化科学与电气工程学院,北京100083

出　　处：《系统仿真学报》2019年第7期1263-1271,共9页Journal of System Simulation

摘　　要：在实际工程中,数据量小、且无评估标准的系统可信度评价问题一直是困扰工程人员的难题。针对该问题,提出了一种将聚类算法和云模型相结合的小样本数据可信度评估方法。利用聚类算法先确定小样本中的聚类中心值,基于此建立云模型。通过云模型产生小样本的扩充数据。根据云滴的置信度分布可进一步计算小样本数据的可信度。将聚类算法与云模型相结合,可以充分挖掘小样本数据中的潜在信息,增加评估的有效性。通过算例分析及仿真证明了所设计方法的合理性和有效性。In practical engineering, the system credibility evaluation with small amount of data and no evaluation criteria has always been a difficult problem for engineers. Aiming at this problem, a small sample data credibility evaluation method combining clustering method and cloud model is proposed. The clustering method is used to calculate the cluster center value for the small sample, and the cloud model is established based on this. The expanded value for small sample data is generated by the cloud model.The credibility of small sample data can be calculated according to the confidence distribution of cloud drops. It can fully exploit the implied information in the small sample data by combining the clustering method with the cloud model, which can increase the effectiveness of the evaluation. The case analysis and simulation are carried out to prove the validity and rationality of the proposed method.

关 键 词：聚类算法 云模型 可信度评估 小样本数据 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]