【作者】 宋良图；

【导师】 葛运建；

【作者基本信息】 中国科学技术大学 ， 模式识别与智能系统， 2007， 博士

【摘要】 生态系统是在一定空间范围内，各生物群落与其环境通过能量流动、物质循环、信息传递而相互作用、相互依存所形成的一个多层次、多因子、多变量的非线性生态学单位。对于生态系统这样的复杂系统，以还原分析为基础的传统建模方法遇到了极大的挑战，仅仅了解系统内部的局部环节而忽略事物间的固有联系，远远不足以解释系统的行为，系统各部分之间的相互作用在解释系统行为方面同等重要。将微观和宏观联系起来的复杂系统理论和基于主体的建模思想为生态系统建模带来了新的启迪。而目前基于复杂系统理论的微观生态系统建模需要解决以下问题1．如何从复杂的微观生态系统中获取基本要素的量化指标；2．如何抽取生态要素之间的互作关系；3．如何为生态过程控制提供仿真试验平台。本文在国家863计划“智能化信息处理技术”课题的支持下，以现代系统理论、方法和现代计算机技术为基础，针对农田尺度的作物生产微观生态系统建模仿真的需要，研究包括微观生态系统数据采集、信息处理、建模仿真和控制决策等理论和技术方法，构建微观生态系统仿真实验平台，为研究微观生态系统行为及其演化规律，提供实验技术支撑。具体研究内容概括如下：1．针对微观生态系统信息采集的特点，利用现代信息技术，设计并实现了一个微观生态系统信息采集平台，平台集成有多种传感器、测量设备和GPRS模块，能够实现微观生态数据，包括土壤、环境以及生物本身的相关信息的实时采集和远程传输。2．研究了生态系统的复杂性，并结合复杂适应系统理论，分析了生态系统的复杂适应性典型特征。在适应性主体概念和复杂适应系统建模方法研究的基础上，结合具体的生态系统实例，提出了微观生态系统的概念模型、表示模型和计算模型，并探讨了复杂性理论在生态学研究中的应用方法。3．将生物中协同进化的概念引入生态系统研究中，研究生态系统内的生物个体、物种、种群、群落的内部，以及它们之间，它们与环境的协同现象、协同进化机理和进化方法，探讨建立生态系统仿真模型的理论基础。在此基础上建立微观生态系统主要生态过程的演进模型。4．研究了适于生态系统建模和仿真的基于Agent的建模思想，以及元胞自动机的工作原理。利用REPAST工具建立了微观生态系统的元胞自动机模型，对生态系统的主要生态过程进行了仿真试验，仿真效果良好。6．重点研究了微观生态系统的可配置仿真平台的系统体系，开发了可以模拟微观生态系统多种行为的仿真模块。通过集成微观生态系统的数据采集、规则抽取、建模仿真、决策控制等功能模块，形成了一种可配置的微观生态系统仿真试验平台。本文研究的部分内容已通过成果鉴定，并获科技部农业成果转化基金的进一步支持，有望通过完善得以实际应用。更多还原

【Abstract】 The ecosystem is a multi-level, multifactor , multivariable and non-linear ecologic unit which covers a certain spatial scope, in which various living things communities and the environment interact with each other through the energy flow, the circulation of materials and the information transmission, and depend on each other mutually. Regarding such complex system, understanding merely the interior partial links in the system is insufficient to explain the system behavior, knowing the interaction of various parts in the system is equally important for the explanation of system behavior, so the theory of complex system and thought of agent based modeling, which bring the microcosmos and macrocosmos together, enlighten the ecosystem modeling. But at present, the following problems should be solved in the ecosystem modeling based on the complex system theory. 1. How to acquire the quantitative indexes of the ecologic factors in the microscopic ecosystem. 2. How to extract the correlations among the ecologic factors. 3. How to provide the ecosystem researchers or administrators with a practical platform for ecosystem modeling and simulation.Based on the modern system theory and its method as well as the modern computer technology, and for meeting the crops production microscopic ecosystem modeling and simulation need, studies including the microscopic ecosystem data acquisition, information processing, modeling and simulation and control decision-making are carried out, and the experiment platform for the microscopic ecosystem modeling and simulation is developed, which can support the study of the microscopic ecosystem. The research results are as followed:1. Microscopic ecosystem information deals with soil, environment as well as living things themselves. When modeling and simulating , many kinds of information should be obtained. According to the characteristics of the microscopic ecosystem information acquisition, and based on the modern information technology, a microscopic ecosystem information acquisition platform is built, which integrates some sensors and the GPRS module, to realize microscopic ecologic data real-time gathering and the long-distance transmission.2. Based on the complexity theory, the concept of adaptive agents and the basic characteristics of complex adaptive system, and from the viewpoint of ecosystem complexity, the ecosystem as a typical complex system is analyzed, and the method of complex adaptive system theory and its application to the ecosystem are discussed.3. In the nature, any species or individuals, any population and communities all exist in a certain ecosystem. In ecosystem, all individuals, species, population and communities compete and cooperate with each other. Study of such coordination and competition relations, is the basis of the ecosystem simulation model, so the paper discusses the co-evolution mechanism and the evolution method of all parts in ecosystem, and explores the ecosystem evolution rules.4. Agent based modeling and simulation method is a new way for complex system modeling and simulation. This paper introduces the Agent-based modeling method suitable for the ecosystem modeling and the simulation, as well as complex adaptive system model frame. It also introduces the structure of Cell-Automation and its mechanism, as well as the CA model of microscopic ecosystem. Meanwhile the paper introduces the platform for the modeling and simulation of multi-agent system.5. At present, a lot of programming work is needed when using complex system modeling tools for ecosystem modeling, so it is difficult for the ordinary user to use the tools .At the same time, there is no a comprehensive tool for the ecologic data acquisition, rule extraction, as well as model validation. The paper focuses on the architecture of the configurable platform of ecosystem simulation, and develops a series of ecosystem models for ecosystem study.Some achievements in this project has been appraised and also funded by agricultural science and technology achievements transformation fund program.更多还原