معايير الضبـط المثلـى للخوا رزميـات الجينيـة للاداء المتصل

Abstract

A class of adaptive search procedures called genetic algorithms (GAs) has been used to optimize a wide variety of complex systems. However, GA's suffer from some disadvantages of identification of the optimal control parameters for GA's are hard. Among certain manipulations are usually essential to speed up and improve the GA performance is optimization of control parameters for genetic algorithms, which must be tuned for efficiency. This work describes experiments that search a parameterized space of GAs defined by seven control parameters in order to identify efficient GAs for the task of optimizing a set of numerical functions. Metalevel GA performs this search. The proposed genetic algorithm has enabled to found optimized genetic algorithm and exceeds over the previous genetic algorithm for the on-line performance only. These optimized algorithms represent a big statistically significant improvement over the expected online performance of the previous optimizing GAS. The results are validated (both the new proposed method & the previous methods) on another set of problems. The results showed the ability and the effectiveness of the proposed genetic algorithm as compared to the optimization rates of the previous GAS.