Dynamic RC operator-based hysteresis model of MR dampers

Abstract

Nonlinear hysteretic mechanical behavior is a major obstacle to the efficient applications of vibration/shock and precision control systems with smart material-based actuators, such as magnetorheological (MR), magnetostrictive, piezoelectric and shape memory alloy actuators. In this paper, a dynamic resistor-capacitor (RC) operator (dRCO)-based hysteresis model (dRCOM) is proposed and thoroughly investigated for precise control of such actuators. The proposed dRCO links the RC operator (RCO) with the input rate, which enables the RCO the dynamic description performance. Employing the principle of graph neural network, the input data is preprocessed by the dRCO, and the neural network model is constructed for the parameter identification of the MR damper model. Experimental tests of the damping force characteristics of a commercial MR damper under different displacement excitations and applied currents are carried out. Correspondingly, mean square errors (MSEs) are compared between the tests and the outputs of the Bouc–Wen model, basic RCO-based hysteresis model (RCOM), and dRCOM. As a result, the dRCOM predicts MR damper more accurately than the other two. Furthermore, simulation and experimental tests of MR damper force tracking using different hysteresis models are also conducted to demonstrate the superiority of the dRCOM.

Adaptive magnetorheological fluid energy absorption systems: a review

“…To accurately describe the dynamic characteristics of MREAs under impact loads and the force characteristics at low speeds in wide ranges of frequency, Bai and Tang (2021) proposed a dynamic RC operator-based hysteresis model. In the RC operator-based hysteresis model, the variable related to the velocity is introduced into the time-varying hysteresis factor, q, to construct the dynamic RC operator, which can be expressed as:…”

“…The neural network models include feed-forward neural networks trained by extreme value learning machines (Bahiuddin et al 2021), neural networks models based on recursive lazy learning methods (Boada et al 2011), and neural networks models using the adaptive differential evolution algorithm to optimize parameters for the support vector regression (Shou 2020). In addition, Bai and Tang (2021) combined a dynamic model based on the RC operator with a neural network model and proposed a dynamic model based on the dynamic RC operator. Parametric models provide clear physical meanings of model parameters, and these non-parametric models show superiority in prediction accuracy.…”

“…where a 1 , a 2 , and a 3 are the constants for the controllable MREA force. The parameters, a 1 , a 2 , a 3 , c 0BH , and F 0 need to be identified in this model, and the main drawback is that the characteristics of the MREA cannot be described well at small deformations and low speeds by this model (Bai and Tang 2021).…”

Magnetorheological semi-active shock mitigation control. Part I: numerical analysis and preliminary tests

“…Mao et al (2013a) also proposed a hydromechanical model based on fluid dynamics, in which the dynamic response of MREA at high velocity is more accurately described by the hydromechanical model than unsteady-Bingham-plastic model and unsteady-BPM model. Bingham model (Stanway et al, 1987), bi-viscous model (Stanway et al, 1996), Bouc-Wen model (Ismail et al, 2009), polynomial model (Choi et al, 2001), resistor-capacitor (RC) operator-based model (Chen et al, 2018), and dynamic RC operator model (Bai and Tang, 2021) are commonly used to characterize the mechanical properties of MREA.…”

Magnetorheological damper current estimation based on k-nearest neighbor algorithm

“…The feed forward, which is derived by a control-oriented mapping approach to reduce modeling effort of the inverse MR damper behavior, compensates for the main steady-state nonlinearity of the MR damper force and thereby linearizes the plant. Recently, Xian-Xu 'Frank' Bai (Bai and Tang, 2021) proposed a dynamic resistor-capacitor (RC) operator (dRCO)-based hysteresis model. The input data is preprocessed by the dRCO, and the neural network model is constructed for the parameter identification of the MR damper model.…”