Land Use Optimization Model Based On Multi-agent System And System Dynamics

Land resources are scarce and non-renewable resources and an important factor in maintaining human survival and long-term stability of ecosystems.However,with the rapid development of the social economy,the contradiction between supply and demand has also emerged.At present,there are widespread problems such as unreasonable land use structure,low land use level,and declining quality of land resources.The optimal allocation of land resource utilization is an important way to promote the intensive use of land resources and achieve sustainable development.Currently,researchers have developed more land use optimization configuration models.However,in general,the adaptive and dynamic control capabilities of these optimized configuration models are weak.the result of optimizing the configuration is also often a two-dimensional feature of quantity and space.The timing process implemented by optimizing the configuration results is not considered.The spatial decision-making behavior between people and land and between people in the land use system is not considered.It is difficult to ensure the rationality and practicability of the model,which makes the land use optimization configuration results more difficult in the process of implementation.Therefore,this paper uses multi-agent system theory to understand and explain the general law of land use change,and combines system dynamics,geographic information science,computer science and intelligent science related theories and techniques to establish a land use optimization model that can clearly express the subject decision（Multi-agent system for Land Use Optimization Allocation,LUOM）.Taking Yinzhou District of Tianjin as the research area,quantitative analysis of regional land use optimization was carried out to explore the internal mechanism and process of regional land use optimization.It provides basic technical support for the formulation of land resource utilization and management policies.The main findings are as follows:1)This study considers the land use system as a complex space system and examines the problem of regional land use optimization from the perspective of complexity.The land use system is simplified into three categories:government agent,department agent（agricultural department,economic department,social department,optimization department）and resident agent（urban land,rural residential area,cultivated land,forest land,grassland and orchard）.The government is a single subject,perceives all the decision-making behaviors of the main body and the macro-level overall situation,and formulates various macro-level多主体系统ter plans,and allocates land resources based on the maximum space efficiency.The departmental agent entity uses the discrete selection model to optimize the location selection,and under the premise of pursuing the assumption of departmental utility maximization,directs the resident agent entity to perform optimization operations.The resident agent entity calculates the suitability value of each candidate position by means of the dynamic random utility model.The variables affecting the multi-agent system include natural variables,environmental variables,location factor variables,traffic accessibility variables,and socio-economic variables.2)In order to achieve better land optimization results,the study used the strategy of sorting and arranging rural residential areas.A rural residential point classification model was established using system dynamics.According to the research situation,the social expansion parameters such as urban expansion rate and population growth rate,the single rural residential areas are divided into urbanized rural residential areas,key development rural residential areas,restricted development rural residential areas and abandoned rural areas.In the multi-agent system,according to the type of rural residential areas,the corresponding types of resident entities are generated,and different optimization strategies are given to different rural residential areas to achieve fine finishing optimal of rural residential areas.3)Among the research results,the regional organic carbon stocks increased from 5.8523million tons in 2015 to 8.421 million tons,an increase of 43.9%,and the total regional economic output increased from 6.981×10¹⁰ yuan to 8.66×10¹⁰ yuan,an increase of 24.1%,the value of land use intensification increased from 0.394 to 0.455,an increase of 15.4%.In the economic priority scenario,compared with the ecological priority scenario,the regional organic carbon stocks decreased to 7,305,500 tons,a decrease of 13%,which was 30%higher than the base year,and the regional economic output increased to 1.09×10¹¹ yuan.At 25.9%,the land use intensification level was 0.54,an increase of 18.7%.The regional organic carbon stocks,total economic output and land use intensification values under the eco-economic coordination scenario were 7,836,200 tons,1.05×10¹¹ yuan and 0.533,respectively,which increased by 33.9%,44.1%and 35.4%.4)In the land use optimization process,the suitability lines of the three scenarios are all oscillating and start to differentiate in the middle of the iteration.The shock response shows the uncertainty and instability of land use optimization in real society,and the rise can also reflect the trend changes brought about by macro-control.The aggregation degree of the three scenarios is a smooth decline in the early stage,and the volatility rises in the later period,mainly because some construction land needs to be demolished in the early stage of optimization,and the demolished land will be repaired later.In the line chart of the landscape pattern index,the plaque density,shape index,and fractal dimension decreased in the early stage and the late stage,and the average area of the plaque increased in the early stage.The shape index polyline trend of the eco-economic coordination scenario is different from the fractal dimension,and the shape index in the final period of the iteration is lower than other scenarios.Considering that the shape index and the fractal dimension respectively represent the similarity of the plaque to the equal area circle and square.That is,in the case of economic priority,the direction of model optimization is closer to the square,so that the plaques based on the grid are more concentrated.In the ecological priority scenario,it is closer to the circle,reducing the edges and corners,easing the damage to the landscape matrix.