Comparação de modelos paramétricos e não paramétricos de atuadores com fluido magneto reológico

Abstract

Since the appearance of semiactive systems in the early seventies, they have been gaining more applications in several engineering projects. In the specific case of semiactive systems using magneto rheological fluid, the first commercial products with this intelligent material were only successfully marketed for the first time in 1998 by LORD® Corporation, the same manufacturer of the actuator used in the present research work. Since then, the applications did not stop growing. From intelligent suspension systems to bridges and buildings, aiming at the safety of people and the structural health of buildings, to vehicle seats, aiming at passenger comfort and safety, magneto rheological fluid actuators occupy a large spectrum of applications. The objective of this work is to present a methodology for the validation of mathematical models, both parametric and non-parametric. For validation purposes, inverse problem techniques were used to optimize the model studied with respect to the experimental data, using the minimization of a relative error, based on the norm of the difference between the forces obtained from the test performed and the numerical model implemented in MATLAB® environment, divided by the norm of the experimental force. The advantages of parametric mathematical models are that they led to a rapid convergence to the results, and the disadvantage is that, since they have a well-defined law of formation, the shapes of the characteristic curves of the actuators are "rigid", i.e., they do not have enough freedom to change their shape drastically. Usually these curves follow a well-defined trend. The nonparametric model studied in this work is based on fuzzy logic, which has a greater "freedom" to model all the points of the experimental curve, conveniently. However, the difficulty in finding the fuzzy parameters is very important, to the point of compromising the validation result. Finally, it was concluded that the parametric hysteretic model presented the best results for design purposes, lower computational cost, and easier implementation as compared with competing models.