The weighted distance between the individual decision results and group decision result is defined to reflect the
consensus of the final decision result, and the adjustment degree of evaluation score matrix is defined to reflect the adjustment to the initial decision preference of decision makers. Under the condition of keeping the initial decision preference, the score matrix is adjusted in order to obtain the new decision result with better consensus. The genetic algorithm(GA) is employed to solve the problem. At the foundation of the score matrix, the coding format of chromosomes is designed. The standard GA is improved by adding the rules to the operations of GA. Finally, the numerical study and the reliability analysis of the result show the effectiveness and reasonability of the proposed method.