LQG Predicted Controller based on RBF neural network

The characters of radial basis function neural network (RBFNN) and linear quadratic Guassian (LQG) algorithm are introduced. For the effect of time lag in the process of structural vibration control, linear quadratic Guassian predicted controller (LQGPC) is presented based on RBFNN. This controller can decrease the influence of time lag availably. A five -story base isolated frame structure is used to evaluate the control effect in the case of time lag and no time lag. Simulation results show that control effect of LQG and LQGPC is equivalent and both two algorithms can decrease the response of the structure effectively in the case of no time lag, but LQGPC still has better ability of controlling structural response in the case of time lag existing in system.