Parameter estimation for the phenomenological model of hysteresis using efficient genetic algorithm
Xue Xiaomin,Zhang Ling,Sun Qing
DOI: https://doi.org/10.1063/1.3452263
2010-01-01
AIP Conference Proceedings
Abstract:Magnetorheological (MR) dampers provide an ideal candidate for diverse applications including structural vibration suppression as well as shock absorption and vibration control in vehicle systems in recent years. To understand the dynamic characteristics of MR damper, many researchers have utilized some useful models to illustrate the behavior of the MR damper, which are Bingham model, polynomial curve-fitting model using sigmoid function, Bouc-Wen phenomenological model, etc. The Bouc-Wen differential model is one of the most widely accepted phenomenological models due to its capability to capture a range of shapes of hysteretic cycles which can match the behavior of a MR damper. As a result, it is the tendency of this model being used to describe hysteretic phenomena. However, this model has an obvious shortcoming, its complicated equation expression, which seems to be a big problem for multi-parameter estimation simultaneously. With the development of the computer technology, a powerful multi-processing algorithm is found out possible to solve these complicated problems efficiently. Thus, this paper devotes to the use of an efficient Genetic Algorithm (GA) to estimate the multi-parameter of the Bouc-Wen model efficiently. A modified GA is adopted and improved by effective methods including adaptive genetic operators and appropriate termination criteria. The modified GA has greatly improved the performance of the traditional GA. Finally, experimental data of the MR damper are used to verify the proposed approaches with satisfactory parameter estimation results and highly efficient computational capability. Meanwhile, the algorithm and the results proposed herein also throw light on development and characterization of other novel smart dampers.