A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Towards an automatic calibration procedure for constrained cellular automata
AbstractThis paper describes an automatic calibration method for the class of cellular automata (CA) characterized by additive transition rules. CA can serve as instruments to support decision-making in spatial systems, since they are in effect dynamic maps representing geographic processes. However, the quality of CA based decision support models depends on the adequacy of the transition rules, which must be complex in realistic models, and these rules typically contain a number of parameters that must be calibrated. At present there is no general method to calibrate the transition rule set other than a manual tuning of the rules. Existing automatic techniques are applicable only to very simple CA. The method presented here draws on several standard optimization and search techniques, adapting them to the CA calibration problem. The procedure is developed first for application to the general class of unconstrained additive transition rules, and is then modified in order to deal with the case of constrained additive rules, where the number of cells in each state at each iteration is determined exogenously to the CA; this latter class is important in urban modeling applications. The calibration procedure is shown to be capable of systematically modifying arbitrary, randomly generated rule sets in order to produce a set of rules that is able to correctly replicate a series of test maps.
Towards an automatic calibration procedure for constrained cellular automata
AbstractThis paper describes an automatic calibration method for the class of cellular automata (CA) characterized by additive transition rules. CA can serve as instruments to support decision-making in spatial systems, since they are in effect dynamic maps representing geographic processes. However, the quality of CA based decision support models depends on the adequacy of the transition rules, which must be complex in realistic models, and these rules typically contain a number of parameters that must be calibrated. At present there is no general method to calibrate the transition rule set other than a manual tuning of the rules. Existing automatic techniques are applicable only to very simple CA. The method presented here draws on several standard optimization and search techniques, adapting them to the CA calibration problem. The procedure is developed first for application to the general class of unconstrained additive transition rules, and is then modified in order to deal with the case of constrained additive rules, where the number of cells in each state at each iteration is determined exogenously to the CA; this latter class is important in urban modeling applications. The calibration procedure is shown to be capable of systematically modifying arbitrary, randomly generated rule sets in order to produce a set of rules that is able to correctly replicate a series of test maps.
Towards an automatic calibration procedure for constrained cellular automata
Straatman, Bas (author) / White, Roger (author) / Engelen, Guy (author)
Computers, Environments and Urban Systems ; 28 ; 149-170
2003-01-01
22 pages
Article (Journal)
Electronic Resource
English
Towards an automatic calibration procedure for constrained cellular automata
| Online Contents | 2004
Calibration of an elasto-plastic constitutive model by a constrained optimisation procedure
| Online Contents | 2006