基于对象增量的不完备混合信息系统动态属性约简算法  

Dynamic Attribute Reduction Algorithm of Incomplete Mixed Information System Based on Object Increment

作　　者：游琪 孙柏杨 YOU Qi;SUN Boyang(College of Computer Engineering Technology,Guangdong Polytechinc of Science and Technology,Zhuhai Guangdong 519090,China;State Grid Hubei Electric Power Co.,Ltd.Xiaogan Power Supply Company,Xiaogan Hubei 432003,China)

机构地区：[1]广东科学技术职业学院计算机工程技术学院,广东广州519090 [2]国网湖北省电力有限公司孝感供电公司,湖北孝感432003

出　　处：《电子器件》2022年第5期1129-1138,共10页Chinese Journal of Electron Devices

基　　金：广东省教育厅特色创新类项目(GXJK320)。

摘　　要：属性约简是粗糙集理论在信息处理和自动化控制领域中的重要应用,然而实际应用环境下,数据的采集是源源不断的并且采集到的数据包含了不同的数据类型,同时数据传输过程中也存在着数据丢失的情况,使得最终得到的信息系统是不完备混合类型的,并且时刻处于增加更新之中。针对这一情形,提出一种条件信息熵的不完备混合型信息系统动态属性约简算法。首先介绍了不完备混合型信息系统的条件熵模型以及条件熵的属性约简。然后提出了不完备混合型信息系统条件熵随论域增加时的增量式更新计算,该计算方法通过旧信息系统的条件熵进一步计算新信息系统的条件熵,理论证明了这种增量式计算具有很高的计算效率。最后基于这种增量式计算,设计出了条件熵的不完备混合型信息系统动态属性约简算法。仿真分析表明,所提出的动态属性约简算法相比较于传统的静态属性约简算法具有很高的动态属性约简性能,同时与同类型的动态属性约简算法相比具有较好的优越性。Attribute reduction is an important application of rough set theory in the field of information processing and automatic control. However, in the practical application environment, the data collection is continuous, and the collected data contains different data types. At the same time, the data loss also exists in the process of data transmission, which makes the final information system is incomplete mixed type, and constantly updated. To solve this problem, a dynamic attribute reduction algorithm based on conditional information entropy is proposed. The conditional entropy model and attribute reduction of incomplete mixed information system are introduced, and then the incremental updating calculation of conditional entropy in incomplete mixed information system with the increase of universe is proposed. In this method, the conditional entropy of the new information system is further calculated by the conditional entropy of the old information system. The theory proves that the incremental calculation has high computational efficiency. Finally, based on the incremental calculation, the dynamic attribute reduction algorithm of incomplete mixed information system based on conditional entropy is designed. Simulation analysis shows that the proposed dynamic attribute reduction algorithm has higher dynamic attribute reduction performance than the traditional static attribute reduction algorithm, and has better advantages compared with the same type of dynamic attribute reduction algorithm.

关 键 词：信息处理 信息系统 数据更新 属性约简 不完备混合型 条件信息熵 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]