Land Use Allocation Optimization Models Based On Multi-Objective Artificial Immune Systems

Land is a very important natural resource, both for survival and development of humanity. However, land resources in China are clearly under tremendous pressure as the country has experienced rapid and sustained urbanization and industrialization. The process of rapid economic development and population growth has intensified the phenomenon of irrational use of land resources, such as land degradation, waste of land resources, and over-exploitation of land resources and etc., which has caused serious ecological and environmental disaster. Therefore, how to use the new theories and methods to guide land-use planning, and to manage land resources efficiently and sustainably has become a hot topic among government, scholars and public in recent years.The main component of land-use planning is land use allocation, which is composed of two types of important problems. The first is to determinate the best land-use structure based on the study of the economic and social development objectives and the land use conditions of the study area. The second is to find the optimal land-use spatial layout under the constraint of the land-use structure and the characteristics of the land resources. To solve the two problems, a lot of models and algorithms have been proposed, such as Mathematical Programming models, System Dynamics models. Cellular Automata models and Intelligent Optimization models and etc. However, most of the models are still cannot meet the demand of solving the land use allocation problems, because most of the them have limitations of poor optimization performance, failed to solve large scale land-use spatial allocation problems or have a bad multi-objective processing strategy and so on. Besides, the models described before were too problem-specific, since scholars merely focused on one specific problem. An intelligent decision support system, which can be used for practical land-use planning, is still unavailable.To advance the models for solving land-use allocation problems and to support the practical land-use planning works, the thesis studies the intelligent land-use allocation model based on the Artificial Immune System, which is an outstanding optimization model and can be defined as intelligent and adaptive computational systems inspired by theoretical immunology principles and mechanisms in order to solve real-world problems. The major works and contributions of this thesis are as follows: Firstly, the objectives and constraints of major problems of land-use allocation are defined clearly under the guidance of the theories of sustainable development, ecological services values, landscape ecology and etc. In the land-use structure optimization problem, maximizing of the ecological services values and maximizing the economic values are considered in building the optimization models. Maximizing the land-use suitability and maximizing the spatial compactness are built into land-use spatial allocation models. The framework of the land-use allocation optimization conceptual model, which is based on the Artificial Immune System, is proposed according to the former definition and the characteristic of the problems.Secondly, the MOAI-LUS algorithm and PAI-LUSA algorithm are proposed for solving the land-use structure optimization problem and land-use spatial allocation problem respectively. To promote the performance of the land-use allocation model, the NICA algorithm, which is an outstanding multi-objective artificial immune algorithm, is adopt to design the MOAI-LUS algorithm and PAI-LUSA algorithm. The initializer operator and mutation operator of the NICA algorithm are re-designed to knowledge-informed operators according to the characteristics of the two types of optimization problems respectively to improve the efficacy and performance of the optimization models. Furthermore, to enhance the capacity for solving large-scale land-use spatial allocation problems, the parallel computing technologies are introduce into the PAI-LUSA algorithm to improve the efficacy of the algorithm and to reduce the optimization time.Thirdly, this thesis describes an intelligent decision support system for supporting practical land-use planning works. The software is developed based on "plugin-host" architecture by the use of C#language and the open-source GIS components called DotSpatial. It provides a series of standard Application Programming Interfaces (APIs) which can:(1) assist researchers in the development of their own problem-specific application plugins to solve different land-use allocation problems; and (2) allow the implementation of some advanced immune operators into the platform to improve the performance of the AIS algorithms. As an integrated, flexible, parallel and convenient tool, the system will advance the development and popularity of AIS based models in land-use allocation.Finally, the optimization of the land-use allocation in Zigui County was chosen as an application to evaluate the performance of the MOAI-LUS algorithm and PAI-LUSA algorithm. Besides, some other outstanding multi-objective optimization algorithms, such as NSGAII, were used for evaluating the performance the algorithms proposed in this thesis. Furthermore, the probability and necessity about the applications of parallel computing technologies to land-use spatial allocation is also evaluated by testing the parallel efficiency of the PAI-LUSA algorithm.Experiments results show that:(1) compared with the NSGAII and some other traditional models, the models proposed in this thesis have consistently demonstrated its outstanding performance.(2) The knowledge-informed immune operators can accelerate the convergence of PAI-LUSA algorithm, as the PAI-LUSA algorithm can obtain land-use spatial allocation solutions with fewer iterations. Furthermore, the parallel efficiency test of the PAI-LUSA algorithm shows that the parallel computing technologies can promote the computational efficiency greatly:when using a computer with16cores CPU to run the optimization task, the algorithm can achieve a quasi-linear speedup of10.83and efficiency of68%. Therefore, it is very significant to using the parallel computing technologies to improve the efficiency of the land-use spatial allocation optimization models especial in solving large-scale problems.