OPTIMIZING DEVICE AND METHOD THEREOF BY GENETIC ALGORITHM BASED ON UNBALANCED EVOLUTION THEORY

摘要

PROBLEM TO BE SOLVED: To eliminate the setting of a mutation rate and to obtain a highly optimum solution from the initial stage of optimization by selecting an optimum solution candi date obtaining maximum adaptability and judging whether the optimum solution candidate is converged to the optimum solution or not. SOLUTION: A candidate group division generation part 4 inputs the first optimum solution candidate from an optimum solution candidate initial generation part 3, repeats division and copy for plural times and generates the candidate group of the optimum solution. An evaluation/selection part 5 calculates optimum degrees on all the optimum solution candidates from an evaluation function inputted from a data condition input part 2 and holds the highest optimum solution candidate to the next generation evolution. A convergence judgment part 6 executes convergence judgment on the selected optimum solution candidate and outputs the optimum solution candidate when the solutions are judged to be converged to the optimum one and the judgment result of the success of convergence to an output part 7. When it is judged that the solution is not converged to the optimum one and the number of generation evolution does not reach the scheduled one, the optimum solution candidate at that time is outputted to the candidate group division generation part 4 and a previous processing is repeated.