A Research For The Theoretical Framework And Modeling Platform For Agent-based Simulation In Geocomputation

From the perspective of human-environment interactions, the dissertation proposes a theoretical framework for agent-based simulation (ABS) in Geocomputation and establishes the corresponding software platform named gis-based agent simulation environment (GBASE). Based on the theoretical framework and software platform, the study on spatial evolution of historical population in China is implemented. The dissertation consists of six chapters according to the structure from overview to theoretical framework exploration to system analysis to system building to application and to conclusion.Chapter 1^st is about the overview of agent-based simulation. On the whole, ABS simulates the objective world based on a series of agents, and the interactions between agents carry out the system dynamics and complexity. It has been one of the new modeling techniques in Geocomputation and has been used for simulating Land Use/Cover changes, urban evolution, residential dynamics, crowd flows, fire spread, tourism and historical geography etc. It is recognized that the wide spread of ABS is partly due to the development of ABS software platforms, such as Swarm, NetLogo, Repast and so on. However, there are still some shortages when those platforms are used in geography, because the development purposes of those platforms are not focused on a specific subject. Especially, the idea of hierarchical modeling which is one of the typical characters in geography is wanting in those platforms. Therefore, it is necessary to develop an ABS theoretical framework and platform for Geocomputation.Chapter 2^nd discusses the theory of human-environment interactions, and propounds the theoretical framework of ABS based on human-environment interactions. From the view of geography, ABS can be considered as the simulation on physical particles which act according to their action rules, and is a typical reflection of the interactions between human and environment. ABS can simulate the complexity of human society from the individuals interactions in micro level and also powerful for testing the validity of the regulations applied in macro level. So the dissertation advances the theoretical framework composed of three main components including geo-automata, individual agent and group agent. Geo-automata is a component for modeling of environment. Individual agent is intended to model the entities with the ability of autonomic decision making. And group agent is a presentation of hierarchy modeling. A group agent on higher level contains a set of individual agents or group agents which belong to the lower levels and so forth to simulate the nested hierarchy. Group agent is designed to satisfy the simulation on hierarchical complex systems.Chapter 3^rd aims to make analysis on the software platform of gis-based agent simulation environment derived from the theoretical framework proposed in chapter 2^nd. By discussing theories of agent-oriented software engineering, the demand analysis of the system is carried out from the view of control construction, script programming and script compiling. To satisfy the system demands, Delphi and FastScript are adopted to develop the platform, and so divide the system into four modules including control constructing module, user script managing module, class library module and script compiling module. And the three main components in theoretical framework are designed into corresponding classes with relevant properties and methods respectively.Chapter 4^th is developed to come through the technical difficulties and introduce the final realization of the platform. By discussing the critical techniques of FastScript, the problems about control construction, script management, operation on GIS data from FastScript and the integration of controls, fastscript and class library are resolved. So GBASE is finally developed. It is a software platform which can used to make further programming in Geocomputation. The main function of the platform includes user control construction, user script programming and class library invoking. Especially, the platform offers a function which can load GIS data handily, so that to simplify the ABS programming in geography.Chapter 5^th is an application on the theoretical framework and GBASE by simulating the spatial evolution of historical population in China. It is a historical population geography model of China by integrating the climatic, agricultural and social influences. In the model, geo-automata is used to simulate environment while individual agents for population, and each province is considered as a group agent. And finally we develop the system on GBASE and make relevant analyses based on scenario hypothesises. The following results are found. 1. It indicates that potential agricultural productivity determined by geographical condition is the essential factor for population distribution in China and works on the initial population distribution as well as the further population development.2.The population in south China surpassed that of north in about 910A.D. mainly droved by AnShi turmoil. 3. The east-west population pattern shaped in about 1240A.D.-1250 A.D. mainly drove by the climate change in 1230-1260A.D.. 4. The impact from climate change seems to be multi-phased, which works weakly before 1230A.D. and becomes much stronger after that. 5. The population center of gravity keeps moving south-east, and the shift in latitude is much more extensive than that in longitude. Finally, based on the application, the shortages of the platform are also discussed.Chapter 6^th is the conclusion and discussion of the dissertation.