Control of the nonlinear building using an optimum inverse TSK model of MR damper based on modified grey wolf optimizer

This paper proposes an inverse model of the magnetorheological (MR) damper, based on a new Takagi-Sugeno-Kang (TSK) model to estimate the required voltage for producing the MR force. Usually, a MATLAB toolbox function, adaptive neuro-fuzzy inference systems (ANFIS), is used to train the TSK models, which uses the gradient-based learning algorithms to tune the weights or membership functions parameters. The main drawback is that, these algorithms most often find themselves trapped in local minima, depending on the initial estimations. To overcome this issue, a grey wolf optimizer (GWO) is selected and modified to achieve the training task and called the model as an optimum modified grey wolf-TSK model (OMGT). Also, the superiority of the modified grey wolf optimizer over its standard one is investigated using some mathematical benchmark test functions. Moreover, the linear quadratic regulator (LQR) controller is designed to estimate the optimal control force of an MR damper. The effectiveness of this optimum inverse model in structural control is illustrated and verified using an eight-story nonlinear benchmark building. The performance of the designed OMGT model is compared with the different control algorithms such on (PON), clipped optimal control (COC), active control, and ANFIS under different earthquakes, which demonstrate an acceptable performance of the OMGT over these control algorithms.