【作者】 蒲海涛；

【导师】 刘法胜；

【作者基本信息】 山东科技大学 ， 控制理论与控制工程， 2011， 博士

【摘要】 智能交互技术是计算机科学和控制科学研究领域中的重要组成部分。随着物联网时代的到来,人类社会、信息空间、物理世界三元世界将实现全面连通与融合,这给物联网环境下的智能交互带来了新的需求。一方面,物联网环境中联网对象来自不同的厂商,可能采用不同的系统平台,导致从这些联网对象上获取的信息结构不一致,从而实现这些对象有效的交互非常困难。另一方面,物联网环境下信息交换和通讯具备多源、异构的特点,交互过程中会出现许多不确定的感知信息,如何获取精确的交互信息,需要研究新方法,提高不确定感知信息的精确性。第三,物理世界的动态性决定了交互环境易变,如何快速准确的适应环境变化引起的交互的变化,给智能交互带来了新的问题。本文利用上下文感知方法研究物联网环境下智能交互问题。首先分析了当前智能交互研究现状及存在的问题,针对传统智能交互的局限性,分析了物联网环境下的智能交互带来的新需求以及上下文感知技术,提出了基于上下文感知的面向物联网的智能交互架构；构造了物联网环境下的智能交互上下文表达模型,利用聚类分析法对原始样本进行离散化处理,然后利用粗集理论对样本决策表进行属性约简,得到能够覆盖原始数据特征的具有最小条件属性的相应学习样本集,再运用神经网络进行模式识别,推理出规则。根据获取的这些规则对交互上下文进行融合,将交互过程中出现的不确定感知信息精确化；基于规则的上下文融合方法,则使得当物理世界发生变化引起交互环境的变化时,只需要修改规则库里的规则就可以适应这些变化,提高了智能交互系统的灵活性。最后,利用本文提出的架构和方法研究智能电网系统模型。结果显示利用本文提出的架构和方法可以很好的实现物联网环境下人机物之间的智能交互。更多还原

【Abstract】 Intelligent Interaction is one of the most important constituent parts of computer science and automatic control. With the advent of the era of Internet of things, the ternary world of the human society, the information space and the physical world will be fully connected and amalgamated. This brings new requirements for the Intelligent Interaction of Internet of things. On one hand, the things of Internet are from different manufacturers, and adopting diverse system platform, it leads to the various structures of the information that acquired from the things, and it is difficult to achieve interaction of them. So a uniform interaction information model of the ternary world of human, computers and things is required to make them capable of recognition and spontaneous interaction with each other. On the other hand, the information exchange and communication of Internet of things are multi-sourced and isomerous, lots of uncertain awareness information appears in the interaction process, and so new method is needed to improve the accuracy of uncertain awareness information. Thirdly, the dynamics of physical world determines the interaction environment is inconstant, and how to adapt to it rapidly, accurately brings a new problem for intelligent interaction.This paper considers the intelligent interaction problem of Internet of things based on context awareness. The current status of the research on intelligent interaction and its existing problem is analyzed, and then a context awareness method to solve the intelligent interaction problem of Internet of things is proposed. The intelligent interaction context description model of Internet of things is constructed. After using the cluster analysis method to discretize the original samples, the attributes of the sample decision table are refined by using rough sets theory to obtain corresponding learning sample sets that cover the original data features with the lowest condition attributes. And then reason out the rules through pattern recognition based on neural networks. According to the rules the interactive contexts are fused to improve the precision of the uncertain sensory information in the process of interaction. Therefore the rule based context is robust to the change of the physical world, and the adaptability of the proposed Intelligent Interaction system of Internet of things is improved.Finally, an intelligent power grid system model is built up with the proposed framework and method. And the context of the interactions is established and made accurate by context fusion. The result shows that the proposed framework and method are effective.更多还原