A.P.C. Cuco, A.J. Silva Neto, F. L. de Souza, H.F. de Campos Velho (2007): Solution of an Inverse Adsorption Problem with an Epidemic Genetic Algorithm and the Generalized Extremal Optimization Algorithm, Inverse Problems, Desing and Optimization Symposium IPDO-2007, April 16-18, Miami (FL), USA.

Abstract: Inverse mass transfer problems have attracted the attention of an increasing number of researchers due to the relevant applications of such problems in the food and pharmaceutical industry. The experimental determination of adsorption isotherms is an important step for the design of new methods in preparative chromatography. The estimation of the adsorption isotherm coefficients using an inverse problem implicit formulation may be performed using either deterministic or stochastic methods. The Generalized Extremal Optimization (GEO) is a brand new evolutionary algorithm, devised to be apllied in complex optimization problems. Based on the Bak-Sneppen simplified model of evolution, it has been applied successfully to design optimization and inverse problems. GEO makes no use of derivatives and can be applied to multimodal or disjoint design spaces, that may have any combination of different types of design variables (continuous, integer and/or, discrete). This makes it very suitable to be used in the problem being tackled here. Simple Genetic Algorithms (GA) use basically three operators: selection, crossover, and mutation. In the present work, a new operator is used, the Epidemical Strategy leading to the Epidemic Genetic Algorithm (EGA). This innovative operator is activated when a specific number of generations is reached without improvement of the best individual. The results obtained with both algorithms, GEO and EGA, presented very good agreement with the experimental data.