A Kind of Adaptive Genetic Algorithm and Application in Nonlinear Model Identification

An adaptive genetic algorithm based on float-encoding is presented and applied in nonlinear model identification. This algorithm is able to modify its own crossover rate and mutation rate during the search according to the fitness adaptively. The improvement can guarantee the colony multiplicity and the convergence. The simulation results of identifying a theoretical model and application to a real object have proved that the adaptive algorithm leads to significantly superior solutions with less computation time.