【作者】 李霞；

【导师】 王煦法；

【作者基本信息】 中国科学技术大学 ， 计算机软件与理论， 2011， 硕士

【摘要】 人体内分泌系统是一个复杂的分布式自适应系统,该系统具有在高层对神经系统的调控作用、功能情感反应、自组织、分布式处理等一系列独特的功能。内分泌系统的上述功能,能够使机体适应各种复杂的内、外环境,有效地维持机体内环境的稳定,使有机体处于最佳状态。抽取生物内分泌系统的信息处理机制,构建体现内分泌机制特点的人工内分泌模型,是人工智能中一项新的重大研究课程。它对人工智能的发展,开发新型的智能信息处理技术及应用,以及复杂问题的求解具有重要的理论意义和应用价值。随着生物内分泌研究的发展,人们开始从分子和细胞的层次来认识荷尔蒙的产生、运输和调节作用,这方面的研究成果也为人工内分泌系统的研究提供了越来越坚实的生理学基础。生物内分泌系统有着与生俱来的自组织能力,可以应用到许多复杂问题的求解中,尤其是针对那些需要在变化的环境中做出决策的问题,因此本文将从不同的方面对人工内分泌系统的模型及应用进行研究,具体来讲,本文的主要工作包括:(1)生物内分泌系统机制挖掘。深入剖析内分泌系统的功能特点及作用机理,对开发新型的智能处理系统及人工智能的发展有重要意义,因此,本文从信息处理角度,对内分泌系统的主要功能和调控机理进行了详细分析,为人工内分泌系统理论模型及应用研究提供了理论基础。(2)基于荷尔蒙的自组织算法及其应用研究。生物内分泌系统通过荷尔蒙的反应扩散过程进行调节,维持机体的内环境稳定,为生物体的正常生命活动提供保障。本文借鉴荷尔蒙反应扩散机制,提出一种基于荷尔蒙的自组织算法,算法中机器人根据接收到的荷尔蒙信息做出行为决策,通过内分泌系统的调节作用,使多个自主体根据环境状态及自身信息及时调整行为,从而提高多自主体系统的自组织能力。为验证算法的有效性,将该算法用于多机器人任务分配中,实验结果表明该方法能够实现机器人的优化分布,即能及时完成任务又可以减少能量消耗。(3)基于内分泌调节机制的遗传算法及其应用研究。本文提出一种基于内分泌调节机制的遗传算法,并用内分泌细胞来模拟种群个体,这些内分泌细胞可以分泌和扩散荷尔蒙,并根据荷尔蒙信息对自身行为作适当调整,从而实现对系统整体功能的调节。为验证算法的有效性,将其用于任务分配问题求解中,实验结果表明该算法具有较好的求解能力。综上所述,抽取内分泌系统的信息处理机制,建立体现内分泌机制特点的智能模型和算法具有重要的理论意义和广阔的应用前景。本文对内分泌系统的信息处理机制进行了初步研究,并探索了内分泌机制在智能控制等问题中的应用研究,如何建立完善人工内分泌模型,探索人工内分泌系统更广阔的应用领域是今后的研究重点。更多还原

【Abstract】 The endocrine system is a distributed and adaptive complex system, which has the properties of regulating mechanism to neural system, emotion reacting mechanism, self-organization, distributed processing and so on. The endocrine system’s properties keep living organism’s life activities and adapt to changes in dynamic environments, both internal and external. Extracting these mechanisms of endocrine system and establishing artificial endocrine model is a newly important research subject of artificial intelligence, which is significant to develop new-style intelligent information processing technology, facilitate the development of artificial intelligence and deal with complex problems.With the study of Biological Endocrinology, people start to understand hormone’s production, transport and regulation from the molecular and cellular level, which provide a more solid physiological basis for the research of artificial endocrine system. Because of the capability of self-organization, endocrine system can be applied to solve many complex problems, especially for those that need to make decisions in the changing environment. This dissertation includes some research in the models of artificial endocrine system and its application from different aspects. The main works can be summarized as follows:(1) Mechanism mining of biological endocrine system. Summarizing the functional characteristics and mechanisms of endocrine system, is significant to exploit new-style intelligent information processing technology and advance development of artificial intelligence. From the information processing point of view, this dissertation summarized the characteristics of regulatory mechanisms in detail. All these provide theoretical basis in research of the models of the artificial endocrine system and its applications.(2) A self-organized algorithm and its application researching based on hormone. Through the reaction-diffusion mechanism of hormones, the endocrine system of organism is maintained homeostasis, and the living organism’s life is kept activities. Inspired by the hormones’reaction-diffusion mechanism, a self-organized algorithm based on hormone is proposed. In this algorithm, through the regulation of the hormones, the robots could adjust their behavior in time and improve the ability of self-organization. A simulation of task allocation in the multi-robot system is performed to verify the validity of the algorithm. Experimental results demonstrate that with the regulatory mechanism, the task can be accomplished effectively, and the topology and routing of the robots are optimized.(3) Genetic algorithm and its application researching based on endocrine system. The genetic algorithm considering endocrine cells as chromosome is proposed in this dissertation. In this algorithm, intelligent agents are viewed as endocrine cells, which can secrete and diffuse hormones. By the regulation of hormones, agents could adjust their behaviors in time to achieve optimization of system performance. In order to prove the validity of this algorithm, a simulation of task allocation is performed, and the experimental results show that the proposed algorithm has superior performance to solve problems.In conclusion, the endocrine system of organism contains special information processing mechanisms, learning from characteristics of the endocrine system, and establishing intelligent models and algorithms have important theoretical significance and wide application foreground. The dissertation investigates some trial research on information processing mechanisms of the endocrine system, and explores their applications in intelligent controls problems. How to create a perfect artificial endocrine model and explore wider areas of application is the focus of our future work.更多还原