New kind of evolutionary strategy and its global convergence properties

On the basis of the algorithm solving the single peak optimization problem, a new evolutionary strategy is put forward. In view of continuous function problems, it is first proved that the (μ+λ)-ES algorithm based on uniform distribution converges in probability in a weaker condition by means of the central limit theorem. Then it is proved that the (μ+λ)-ES algorithm which adopts the general continuous random variables as its variable operators converges in probability. Numerical results indicate that adopting the evolutionary strategy based on uniform distribution to solve the optimization problems of the continuous function with comparatively higher dimensions can converge to its global optimum solution quickly and effectively.