粗糙集数据分析系统的程序实现

Program Realization of Rough Set Data Analysis Systems

粗糙集理论是一种新的处理不精确、不完全与不相容知识的数学工具。首先简述基于粗糙集方法的数据分析系统的基本构成,分析了粗糙集中连续属性离散化的方法,实现了连续属性数据进行数据离散化。围绕不可区分关系和相对正区域两个核心概念,通过知识之间的依赖程度,提出了粗糙集数据分析的算法,通过比较属性约简的数目,选择最少属性数量的约简结果。得到了求取相对核、上(下)近似集、等价关系、相对重要度、属性相对约简、范畴相对约简、最小决策规则等的各种算法的程序实现。给出了利用MATLAB实现该系统约简化、核及最小决策规则的程序。最后给出实际工程系统的程序运行结果,对滚动轴承故障诊断的仿真实例表明,该方法简化了诊断规则,得到较高的故障诊断正确率。对推动粗糙集理论在具体实践中应用具有实际意义。

Rough set theory is a new mathematical tool to deal with imprecise,incomplete and inconsistent data. First, the basic constitute of data analysis system based on rough set method was briefly described. Discretization method of continuous attributes was considered and continuous attributes were changed into discrete attributes. Two important concepts of indiscernibility relation and relatively positive region were mainly focused on. Using the dependant degree of knowledge, the algorithm of rough set data analysis system was submitted. Comparing the numbers of reduced attributes, the result of minimal attributes reduction was selected. Program realization of many algorithms of solving relative core, upper(lower) approximation, equivalence relation, relatively significant degree, relatively attributes reduction, relatively value reduction, minimal decision rules was obtained. The MATLAB programs of above fields were given. At last, running results of factual engineering system were promoted. Simulation results for rolling bearings show that the method improves the rate of fault diagnostic, simplifies the diagnostic rules. It is obviously factual significant in promoting application of rough set theory.