Desarrollo de dos modelos inversos de un amortiguador magneto-reológico para el control de vibraciones en estructuras civiles

this paper presents the development of two models that emulate the inverse dynamic of a magnetorheological damper, these models estimate the required voltage to produce the force determinated by some linear control strategy. the first inverse model has been implemented using a multilayer perceptron neural network trained under the levenberg-marquardt？s algorithm, and the second model is based on a fuzzy identification strategy, it consists in a first order takagi-sugeno model whose rules are created by the fuzzy c-means clustering method. finally, and as validation, it evaluates the behavior of these models to reduce the seismic response of a three degrees of freedom building using a linear quadratic regulator. in the conclusions is discussed how successful results incorporate these models in a linear control loop