Abstract: An improved method(GA +) for feed formulation design is presented.Computational efficiency is increased by improving and optimizing the traditional genetic algorithm selection strategy,search space,adaptive crossover rate and mutation rate,and adopting excellent individual group to reduce the search space in genetic evolution process.Application of the method to emulation experiment with three classic nonlinear functions,it shows that the convergence speed and global optimization are better than the ones in traditional genetic algorithm.The result shows that GA + does well in maintaining the population diversity,high precision,fast convergence speed,and is very effective in solving design problems of feed formulation.