The invention relates to a global group optimization method based on an optimal individual Lipschitz supporting face. Under the basic difference evolution algorithm framework, on the basis of a Lipschitz supporting idea, firstly, a Lipschitz estimation lower bound supporting face for each individual in an initial group is established to form initial lower bound estimation of a target function; secondly, whether target function evaluation on a new indivisual is worth doing is determined according to lower bound estimation information of the acquired new individual, and a partial invalid area is found out by utilizing extremum information of a lower bound estimation area; thirdly, local increase is realized by utilizing a generalized decrease direction of the Lipschitz lower bound supporting face, and an algorithmic convergence speed is accelerated; fourthly, a Lipschitz lower bound supporting face for the optimal individual in a group is established, and lower bound estimation is made to continuously approach the target function. The global group optimization method has advantages of relatively high reliability, relatively low calculation cost and relatively rapid convergence speed.