A new fuzzy genetic algorithm based on population diversity

Proposes a kind of population diversity (PD) measurements and their computational formulas. A quantitative relation between them is proved. The gene and individual are considered as separate units to investigate PD. A fuzzy genetic algorithm is designed using PD measurements developed, in which the fuzzy controller is used to adjust crossover rate and mutation rate dynamically to maintain the proper PD during the GA´s operation. Experiments prove that premature convergence can be overcome effectively by controlling PD during the GA´s operation.