Subject On Modified Hybrid Genetic Algorithm

In order to overcome the slower convergence of the standard genetic algorithm (SGA), and in which the local search is weak, the paper will combine the traditional optimization and genetic algorithm. First by introducing the sufficient decrease condition,the Wolfe condition and the Goldstein condition with their strong local search capabilities, do the step length selection. The search direction is using conjugate gradient direction. An efficient and convergent hybrid genetic algorithm is obtained.In order to overcome the limitations of the Golden selection method in solving the Multi-optimization problems, in accordance with its outstanding performance in a single optimization problem, inspired by the one-dimensional golden selection, in the literatures [27] [28] [29], 2D golden selection algorithm is extended to 3D, and then extended to high-dimensional, thus gained a hybrid genetic algorithm deformation with good numerical results. Through the baseline function test, we compared the numerical results of these algorithms and knew the strengths and weaknesses of Algorithm.To solve the constraint optimization problems , the hybrid genetic algorithm with the filter method is a kind of using the filter method, and is different from dealing with penalty function, and different from the multi-objective planning new hybrid genetic algorithm.