In recent years, with the maximization of control design and efficiency, and the improvement of economy and energy efficiency, building technology and control in the theory have attracted the attention of lots researchers. By trying various control theorems, many numerical methods have been investigated in the literature to achieve this target, but all these numerical methods are difficult to work out the problem correctly. This paper puts forward a potentially feasible evolutionary bat algorithm (EB) method for active control of earthquake-induced vibration in building structures. Based disturbance observer based control and S surface combined with the robust adaptive control scheme for solving optimization problems proposed, an important contribution in the control law is what the configuration control in the present study should not require known uncertainty limits and the disturbance is eliminated. A simulation case study was proposed to illustrate the possibility of implementing an apparent learning method in ANN to effectively control structural vibration under the influence of systematic motion under earthquake citations. The proposed learning numerical methods does not need to develop a mathematical model of structural dynamics or train another neural network to approximate the actual structural response to be performed.