多粒度双量化决策粗糙集及其属性约简研究

发布时间：2018-05-31 07:52

本文选题：广义多粒度 + 双量化　；参考：《重庆理工大学》2017年硕士论文

【摘要】：随着科学技术的迅猛发展,在科学领域、经济领域及社会生活的方方面面都出现海量数据,这些数据具有信息量巨大、类型繁多、价值密度低、处理速度快等特点。如何快速、高效地从许多信源搜集到的庞大数据中获得有价值信息不仅是信息技术研究的热点,也是目前人工智能领域所面临的巨大机遇与挑战。在越来越复杂的数据环境中,需提出更多的具有针对性的数据处理模式,以便从数据中发现隐含知识,揭示潜在规律。Pawlak粗糙集理论是基于不可区分关系建立分类机制,通过上下近似算子刻画不确定性信息。自1982年提出已经被广泛证实是高效的表达和处理各种不完备信息的数学工具。随着研究的深入,大量的推广模型被提出。本文基于变精度粗糙集和程度粗糙集,先研究广义多粒度双量化决策粗糙集理论,其次研究序信息系统、直觉模糊信息系统的“逻辑或”双量化粗糙集理论,最后为消除冗余信息对计算过程和最终结果造成的影响,研究多源信息的属性约简。主要创新点如下:1.基于少数服从多数决策原则和双量化决策的容错能力,研究广义多粒度双量化决策粗糙集理论。通过定义上、下支持特征函数给出两型广义多粒度双量化决策粗糙集的上下近似算子,并研究两型决策模型的重要性质和决策规则、一定约束条件下两型决策模型的关系以及两型粗糙集模型与其他粗糙集模型的关系。最后通过案例充分展示两型粗糙集模型的分类优势.广义多粒度双量化决策理论为决策理论、多源信息融合和广义多粒度粗糙集的推广提供了理论基础。2.通过变精度与程度“逻辑或”双量化指标研究序信息系统的粗糙集理论。先在序信息系统中研究一个模糊概念的“逻辑或”双量化近似刻画,提出变精度与程度“逻辑或”粗糙模糊集;后通过对象关于属性的加权得分函数在直觉模糊系统中定义优势关系,提出直觉模糊序信息系统下的变精度与程度“逻辑或”粗糙集。同时,研究了两个所提模型的基本结构与重要性质。最后通过案例验证了模型的合理性、有效性以及可行性。3.在多源决策系统中,基于原始有效信息的完全保留定义多源决策系统的一致属性约简,同时为增强实际生产环境的适用性,基于原始有效信息部分保留提出条件熵融合的属性约简;在多源模糊决策系统中基于模糊粗糙集理论通过最大最小、汉明和欧几里得贴近度定义的模糊相似度研究属性约简。最后,通过案例深入阐述多源决策和多源模糊决策系统的属性约简理论,为粗糙集模型的属性约简提供了理论基础。
[Abstract]:With the rapid development of science and technology, huge amounts of data appear in the fields of science, economy and all aspects of social life. These data have the characteristics of huge amount of information, various types, low value density, fast processing speed and so on. How to quickly and efficiently obtain valuable information from the huge data collected from many information sources is not only a hot topic in information technology research, but also a great opportunity and challenge in the field of artificial intelligence. In the increasingly complex data environment, more and more targeted data processing patterns should be put forward in order to discover the hidden knowledge from the data, and to reveal that the latent rule. Pawlak rough set theory is based on the indiscernibility relation to establish the classification mechanism. Uncertainty information is characterized by upper and lower approximation operators. Since 1982, it has been widely proved to be an efficient mathematical tool to express and process all kinds of incomplete information. With the development of research, a large number of generalized models have been proposed. Based on variable precision rough set and degree rough set, this paper first studies the generalized multi-granularity double quantization decision rough set theory, then studies the "logic or" double quantization rough set theory of order information system and intuitionistic fuzzy information system. Finally, in order to eliminate the influence of redundant information on the calculation process and the final result, the attribute reduction of multi-source information is studied. The main innovations are as follows: 1. Based on the principle of majority decision and the fault-tolerant ability of double quantization decision, the rough set theory of generalized multi granularity double quantization decision making is studied. The upper and lower approximation operators of two types of generalized multi-granularity double quantization decision making rough sets are given by the lower support characteristic function, and the important properties and decision rules of the two types decision models are studied. The relationship between two types of decision models and between two types of rough set models and other rough set models under certain constraints. Finally, the classification advantages of the two types of rough set model are fully demonstrated by a case study. The generalized multi-granularity double quantization decision theory provides the theoretical basis for decision theory, multi-source information fusion and generalized multi-granularity rough set. The rough set theory of order information system is studied by variable precision and degree logic or double quantization index. Firstly, the "logic or" double quantization approximation of a fuzzy concept is studied in the order information system, and the variable precision and degree logic or "rough fuzzy set" is proposed. After that, the dominant relation is defined in the intuitionistic fuzzy system by the weighted score function of the attributes, and the variable precision and degree logic or "rough set" under the intuitionistic fuzzy order information system is proposed. At the same time, the basic structure and important properties of the two proposed models are studied. Finally, the rationality, validity and feasibility of the model are verified by a case study. In the multi-source decision system, the consistent attribute reduction of the multi-source decision system is defined based on the original valid information, and in order to enhance the applicability of the actual production environment, The attribute reduction of conditional entropy fusion is proposed based on the original effective information, and the attribute reduction is studied based on the maximum and minimum fuzzy rough set theory and the definition of similarity degree between hamming and Euclidean in multi-source fuzzy decision-making system. Finally, the theory of attribute reduction for multi-source decision making and multi-source fuzzy decision system is expounded through case studies, which provides a theoretical basis for attribute reduction of rough set model.
【学位授予单位】：重庆理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【参考文献】