Optimization of control parameters for genetic algorithms
IEEE Transactions on Systems, Man and Cybernetics
Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence
Modeling data envelopment analysis by chance method in hybrid uncertain environments
Mathematics and Computers in Simulation
| Hi-index | 0.00 |
A new approach for the source quantification has been developed on the basis of real air pollutant hourly concentrations of SO"2, measured by three monitoring stations, during 9h, around a group of three industrial sources. This inverse problem has been solved by coupling a direct model of diffusion (Pasquill's Gaussian model) with a genetic algorithm, to search solutions leading to a minimum error between model outputs and measurements. The inversion performance depends on the relationship between the wind field and the configuration sources-receptors: good results are obtained when the monitoring stations are downwind from the sources, and in these cases, the order of magnitude of emissions is retrieved, sometimes with less than 10% error for at least two sources; there are some configurations (wind direction versus source and receptor locations) which do not permit to restore emissions. The latter situations reveal the need to conceive a specific network of sensors, taking into account the source locations and the most frequent weather patterns.