基于关联系数靶心距的混合多属性识别

Hybrid multiple attribute recognition based on coefficient of incidence bull's-eye-distance

针对目标的实数型、区间型和序列类型的混合多属性数据与数据库中目标的区间多属性数据的识别决策问题,提出了基于关联系数靶心距的混合多属性识别方法。定义了混合多属性数据的度量方法,提出了一种新的灰靶形成规则,以目标混合多属性数据与数据库区间数据的关联系数作为识别决策矩阵,以此形成正负靶心,计算各模式的正负靶心距,讨论了现有靶心距决策的不足,结合一种新的靶心距识别决策方法进行识别判定。结合混合多属性数据识别仿真实验,验证了所提出的新的灰靶识别方法的有效性,并将各种靶心决策方法、灰关联识别法与本文方法进行了性能对比,突出了本文方法识别决策区分度好、稳定性高的优点。

A hybrid multiple attribute recognition method based on coefficient of incidence bull＇s-eye-distance is proposed for the recognition and decision-making problem, that is, hybrid multiple attribute data of the real, interval and sequence number could not be recognized with the interval data in the database directly. The measurements for the hybrid multiple attribute data are determined and a novel grey target forming framework is presented, in which we use the coefficient of incidence of the hybrid multiple attribute data and the interval data as the recognition and decision-making matrix and the positive and negative clouts are formed based on it. Then we calculate distance of the each mode and the positive and negative clouts. Finally we discuss the shortcomings of the existing grey target decision-making methods and present a novel bull＇ seye decision-making method for recognition with the bull＇ s-eye-distance. With the multiple attribute recognition simulation experiment, we show the effectiveness of the proposed novel grey target recognition method and compare it with bull ＇s-eye decision-making methods and the grey association recognition method, which underlines the good distinguishing performance and stability in recognition and decision-making.