基于加权的核事故聚类分析

核事故对人类的生产生活会造成极其严重的后果.对其进行科学分类是安全管理和核事故处理的重要基础.分析了核事故造成的死亡人数、有害照射总剂量、受照人数、经济损失等多属性特征;通过灰关联度确定了各属性指标的不同权重,定义了事故间的加权指数相似度;根据事故属性指标的相似度,利用基于更为科学的进化聚类分析对事故进行了分类;并以近来全球发生的6起重大事故为例进行了实证分析.提出的事故分类方法,综合考虑了事故造成危害的机理,理论严谨、方法实用有效,可作为有关部门进行核事故分类和核事故处理的决策依据.

Cluster Analysis of Nuclear Accidents Based on Weight

Nuclear accidents do great harm to our human's life and working. So it is a very important basement for safety management and disposing those accidents after they occurred to categorize those accidents scientifically. Multiple characteristic attributes such as death, total toxic dosage, total irradiated people or economic loss, are analyzed in this thesis. Different weights of each attributes are determined by degree of gray correlation. The exponential similarities of those accidents are defined. According to those similarities and a more scientific method, those accidents are categorized by clustering analysis. Taking 6 severe accidents occurred in 20 years as examples, the analysis is exemplified. This thesis puts forward the method for categori- zing accidents, which synthetically calculates the harms of those accidents. The method is feasible and effective, and could be used as the criteria in decision-making in categorizing and disposing of nuclear accidents.