To solve the optimization problem of control algorithm parameters， damper parameters， and layout location in the semi-active control systems with magnetorheological （MR） dampers， an improved adaptive niche genetic algorithm is proposed. The proposed genetic algorithm is improved in selection strategy， crossover and mutation operation， and adaptive adjustment of crossover probability and mutation probability. Two different niche technologies， the pre-selection mechanism and sharing mechanism， are used in the improved genetic algorithm. The results of a numerical example show that the optimization result of the improved adaptive niche genetic algorithm （GA-Ⅰ） and an improved simple genetic algorithm （GA-Ⅱ） is generally consistent， indicating the correctness of the former algorithm. Moreover， the GA-Ⅰ consumes an average of 32.7% less computation time to obtain the optimal solution for the first time than the GA-Ⅱ， which means that the former algorithm converges faster than the latter. In addition， optimization results of 30 times indicate that the GA-Ⅰ has stronger stability than the GA-Ⅱ. The semi-active control system with MR dampers optimized by the GA-Ⅰ achieves effective vibration suppression. The maximum values of inter-story drift angles and floor absolute accelerations of the semi-actively controlled structure under El Centro， Chi-Chi and man-made waves are decreased by 64.1%， 54.7%， and 55.9% on average compared to those without control， respectively. The numerical example demonstrates the effectiveness of the GA?Ⅰ， and the integrated optimization of the semi-active control system with MR dampers is realized.