최적 용도지역 배치를 위한 GA 시뮬레이션

GA 시뮬레이션으로 최적 용도지역 배치 방안을 모색했다.

In urban planning, land use arrangement problem is essential and important. Until now, land use arrangement depends on a planner's rational thought process. The planner tries to find better land use pattern in a given situation by many sided analysis, comparison and examination. However, as the value considered in land use arrangement get more and the correlation among values gets more complex, it is not easy to find the best land use pattern. The development of computer systems and various analysis techniques using it help solve complex urban problems.

Genetic algorithm(GA), one of the techniques of artificial intelligence modeling the phenomenon of inheritance and selection, is being studied in the areas of architecture and urban planning. GA has several advantages. Most of all, it is an analysis technique most prominent in finding Global Optimum. This study aims at constructing a model of land use arrangement using GA and developing a PSS(planning support system) capable of simulating according to the variation of conditions. First, urban regions were divided by grid in order to apply GA to arrangement problem and chromosome code method was studied.

This study try to the capability of applying GA to land use arrangement, selecting both location appropriateness and combination considering adjacency of cells as a variable. GA model constructed Delphi 5.0 language. Second, constructed GA model was applied to case areas which are Bang-eo dong, Hwa-jeong dong and Il-san dong in Ulsan. Data construction and simulation result was visualized by linking ArcView GIS of Esri Co. The simulation is an association of each parameter, case 1 considering location appropriateness only, case 2 combination only, case 3 both location appropriateness and combination.

The values of GA parameters were first population size 100, 200, 300, crossover number point 1, 2, 3 and mutation rate 0.01, 0.03, 0.05 and the simulation was carried out with the association of values of each parameter. The simulation result verified how well land use arrangement was reappeared, which is randomly arranged now. According to the simulation result, GA model constructed through this thesis reappeared present land use pattern, which was randomly arranged, with accuracy of case 1 0.9, case 3 0.95 and more. And Users were free to set up target areas, the values of each parameter and GA parameter in interface constructed in Delphi 5.0.

And the simulation result was visualized by linking ArcView GIS of Esri Co. PSS for land use arrangement which has been developed through this thesis was proved to carry out simulation effectively about various scenarios.