摘要
针对无监督情况下的电子产品健康聚类问题,提出一种改进的免疫遗传模糊C均值(IGA-FCM)聚类模型。综合利用多参数历史信息,通过引入加权相似度度量,刻画不同参数对健康状态的影响程度;通过将免疫机理引入到遗传框架中,以FCM的目标函数为搜索因子,克服FCM算法对初始中心选择敏感及遗传算法的早熟等问题。实验结果表明,该模型具有较高的收敛精度、收敛速度和对对象的刻画能力。
According to the problem of health condition cluster for electronic products under the unsupervised case, proposed a model based on modified Immune Genetic Algorithm and Fuzzy C-Means （IGA-FCM）. Compre- hensively utilizing of multi-parameter history information, by introducing of weighted similarity measure, character- ized the impact state of different parameters on the health condition. Through introducing the immune mechanism into the genetic framework and transforming the objective function of FCM into search factor, overcame the problems of FCM algorithm which is sensitive to initial center and genetic algorithm which is easy to be precocious. The ex- perimental results showed that the model has higher convergence precision, speed and abilities to describe the ob- ject.
出处
《科学技术与工程》
北大核心
2013年第16期4585-4590,共6页
Science Technology and Engineering
基金
总装武器装备预研基金项目(9140A27020212JB14311)资助
关键词
健康状态聚类
电子产品
加权相似度
免疫遗传
模糊C均值
health condition cluster electronic product weighted similarity measure immune genet-ic algorithm fuzzy C-means
