A new neural network (NN) predictive controller (NNPC) algorithm has been developed and tested in the computer simulation of active control of nonlinear benchmark buildings. Although in classical model predictive control (MPC) usually a linear model of structure is used, NNPC provides a nonlinear model. In the present method an NN is used as an emulator. This emulator NN has been trained to predict the future response of the structure. The trained NN provides a model of structure which is employed to determine the control force via a numerical minimization algorithm. Since the NNPC controller is very time consuming and not suitable for real-time control, it is then used to train a NN controller. The approach is validated by using simulated response of a nonlinear benchmark building excited by several historical earthquake records. The simulation results are then compared with a linear quadratic Gaussian (LQG) active controller. The results indicate that the proposed algorithm is effective in relative displacement reduction which is here selected for control.

Keywords