Study On Method And Application Of Land Use Spatial Clustering Based On Intelligent Computing Technology

With the development of science and technology spatially"3S", the acquiring ability of land use information has been enormously enhancing and the knowledge demand of land resources management has been gradually increasing at meanwhile, and it is an effective approach for spatial data mining to resolve the current contradiction in land use information.As an important method of spatial data mining, the spatial clustering applies widely in land use data mining. However, at present the algorithm and the realization process of the spatial clustering mining have limitations, and lack of intelligence, dynamic and agility in some degree. The application of intelligent computing model could resolve the shortages effectively.The study was taken from three aspects: theory, technology and demonstrations of spatial clustering. As to the land use problem, through the combination of the intelligent computing method and traditional spatial clustering method, the paper tried to achieve the improvement of genetic algorithm and neural networks to the spatial clustering algorithm, and the optimization of spatial clustering model based on GIS. And then the model was applied in the two land use demonstrations, the first was classification of the urban land, and the second one was prime farmland designation, which include both rural and urban regions, were involved with the problems of urban land grading and cultivated land protecting, and two examples were representative and comprehensive. Finally, significative conclusions were acquired. From another aspect, this paper also can be considered to be a discussion of finding rules and mining knowledge in land use data, and the spatial data mining is the right method to realize that.First of all, on the aspect of theory, the paper brought forward the main tasks of land use spatial clustering mining, then analyzed the key problems on the spatial clustering, and discussed the spatial data type, the cluster type, and the similitude measurement and so on, which extended clustering into spatial clustering.Secondly, as to the technology of spatial clustering, the paper proposed new computational formula of the weighted vicinity degree, and achieved improvement of the spatial clustering method based on intelligent computing model. Two tasks were developed: one was to achieve improvement of K-Means in the spatial clustering by genetic algorithm, the other one was to realize the spatial clustering by self-organizing map. Finally, we designed and achieved respectively the spatial clustering system framework based on VB, MapX and Matlab platforms.Finally, demonstration studies were done. The first one was about the mining of land use spatial differentiation, taking the urban land grading for example, which achieved improvement and optimization of both the multifactor integrative classification and land price divisional classification, and designed respectively the improved technical route of the spatial clustering for two methods. Through the demonstration study of partial regions in Jinan, the result indicated, the method of the urban multifactor integrative classification based on the spatial clustering had higher distinguish ability, and could wipe off evidently the fragmentized shivers; the land price divisional classification based on the spatial clustering could achieve the land price district grading rapidly and scientifically in the land market developing regions, and realize the grading fleetly and effectively. The second one was the demonstration study of the land use spatial structure optimization, taking prime farmland grading application for example, which brought forward the improved project and technical route based on the spatial clustering. Through the demonstration study of prime farmland grading in Jiyang County, the result indicated that prime farmland based on the improved spatial clustering aggregated distribution obviously, the variance monished evidently in this region, which wiped off the isolated plot fragments availably, was in favor of protecting, managing and measuring dynamically prime farmland, and forced the scientific and scale management of prime farmland.In conclusion, the paper summarized main content, brought forward the aspects that need to be improved, and putted forward the study expectation in the future.