Neuro-fuzzy control strategy for earthquake-excited nonlinear magnetorheological structures

Magnetorheological (MR) damper is a prominent semi-active control device for earthquake responses mitigation of structures. The most important topic for the intelligent MR structures is choosing the control current of MR dampers quickly and accurately. The typical control strategy is on–off control strategy, i.e. bi-state control strategy, while inherent time-delay and coarse control precision lie in this strategy. This paper proposes neuro-fuzzy control strategy, in which the neural-network technique is adopted to solve time-delay problem and the fuzzy controller is used to determine the control current of MR dampers quickly and accurately. Through comparison between the bi-state control and the neuro-fuzzy control strategies and a numerical example about a three-story reinforced concrete structure, it can be concluded that the control strategy is very important for semi-active control, the neuro-fuzzy control strategy can determine currents of MR dampers quickly and accurately, and the control effect of the neuro-fuzzy control strategy is better than that of the bi-state control strategy.