【作者】 王桂平；

【导师】 李祖枢；

【作者基本信息】 重庆大学 ， 模式识别与智能系统， 2003， 硕士

【摘要】 控制理论经过100多年的发展，在描述相对简单的系统方面取得了巨大的成功。然而，现实生活中更多的是复杂系统，如复杂的工业生产过程、计算机集成制造系统、智能机器人系统等，控制理论亟需在复杂系统控制领域取得突破性的进展。与此同时，复杂性科学在对复杂系统的描述与模拟，并将其应用于探讨复杂系统的进化与功能的本质方面取得了显著的成功，但在控制这些系统方面还无能为力。因此控制理论与复杂性科学正越来越趋向于对同一问题，即复杂系统控制的研究。正是基于上述背景，本文在分析图式理论和仿人智能控制理论的基础上建立适合于复杂系统控制的仿人智能控制理论——基于动觉智能图式的仿人智能控制理论(SMIS-HSIC)。本文主要完成了以下工作：(1) 综述认知心理学和人工智能等领域的图式理论体系，在综合其他学者对图式概念理解的基础上，给出了图式更详细更精确的定义，这对本课题的后续研究起了规范性、指导性的作用。(2) 探讨图式理论应用于复杂系统控制的可能性，分析其研究内容和关键问题。从图式理论和人体的动觉智能结合的角度研究动觉智能图式。分析人体控制系统的结构，提出从动觉智能图式的角度模仿人体控制系统的思想。(3) 分析机器人控制结构设计中应用很广并比较成熟的方法——运动图式法，这对定性定量描述图式具有着重要的借鉴作用。(4) 介绍图式学习和进化的基础理论——遗传算法，给出了遗传算法中图式的定义、相关的几个概念以及图式定理。动觉智能图式的同化、顺应和学习进化的方法和算法研究是本课题后续研究的主要内容之一。(5) 本文的重点在于讨论建立基于动觉智能图式的仿人智能控制理论。研究感知图式、运动图式、关联图式以及动觉智能图式的结构和描述方法，并给出了它们的详细定义。讨论仿人智能单元控制级中的动觉智能图式思想，探讨基于动觉智能图式的仿人智能控制理论的基本思想、亟需解决的难题，探讨基于动觉智能图式的仿人智能控制器设计方法，并以小车—二级摆的摆起倒立控制为例加以阐述。更多还原

【Abstract】 Having been developed for over 100 years, control theory has made a great success in describing relative simple systems. But there are more complex systems in real life, such as complex industrial production processes, computer integrated manufacturing systems, intelligent robot systems, etc. It is urgent for control theory to make a remarkable progress in the control of complex systems. At the same time, complexity science has made a success in describing, simulating complex systems and applying these achievements in discussing evolution of complex systems and their essence of function, but it is incapable of controlling these complex systems. Therefore control theory and complexity science tend to the research of the same problem, that is, control of complex systems.Under this background and based on analyzing schema theory and Human Simulated Intelligent Control(HSIC) theory in detail, we try to establish HSIC theory fit for the control of complex systems, that is, Sensory-Motor Intelligent Schemas based HSIC(SMIS-HSIC). The main works of this dissertation are shown as follows:(1) We summarize schema theory in cognitive science and artificial intelligence. Based on the comprehension of other researchers, we put forward a preciser and more detailed definition of schema, which is of instructional and canonical significance to the successive researches of this project.(2) We discuss the possibilities, research contents and key problems of applying schema theory to the control of complex systems. We study SMIS from point of view of combining schema with kinesthetic intelligence, that is, Bodily Kinesthetic Intelligent Schema(BKIS). We analyze architecture of human body control system, and put forward the thought of imitating human body control system from BKIS point of view.(3) We analyze Motor Schema method, which is a ralative mature technology and widely used in design of control architecture of robots, and from which we can distill the thoughts in qualitatively and quantificationally describing schemas.(4) We introduce basic theory of schemas in learning and evolution, that is, genetic algorithm. We discuss the definition of schemas, several relative concepts and schema theorem in genetic algorithm. Algorithms research of SMIS in assimilation, accommodation and learning is one of the main works of successive researches.(5) The emphasis of this dissertation lies in establishing SMIS-based HSIC, including describing perceptual schemas, motor schemas，associated schemas and SMIS qualitatively and quantificationally, giving their structures and detailed definitions, discussing basic thoughts, difficult problems of SMIS based HSIC theory, discussing design steps of SMIS-based HSIC controller, etc. At last, we give an example, that is,<WP=6>swinging up and handstand control of cart-double pendulum system, to illuminate design method of SMIS based HSIC controller.更多还原