본문 바로가기
  • Home

Calibration of Dynamic Spatial Model Using Genetic Algorithms

김복환 1 Kwang-Sik Yang 2

1국토해양부
2순천향대학교

Accredited

ABSTRACT

The attraction of Cellular Automata(CA) urban growth models resides in their simplicity, flexibility, and transparency as a modelling framework, and they have been cited in much of the literature as a promising tool to model realistic urban growth. However, the lack of a rigorous calibration process for CA urban growth models is certainly one potential obstacle to their use. Researchers have found the solution for the calibration problems of CA urban growth model from the field of Artificial Intelligence(AI) equipped with heuristic search routines. AI technique of Genetic Algorithms(GA) can be one of the promising alternatives. A model for new city growth (the so-called NCGM: New City Growth Model) was developed using stochastically constrained CA. A GA was designed to calibrate the NCGM. Accuracy and consistency of the calibration results through the GA were sought using hypothetical data and real data and meanings for the calibrated values are sought. It was revealed that NCGM produced very reliable results based on the experiments using both hypothetical and real data. Underpinned by GA’s accurate and consistent calibration results CA urban growth models, such as NCGM, can be strengthened their position in simulating realistic urban growth.

Citation status

* References for papers published after 2023 are currently being built.