Contents I. Introduction
The suspension system is the connection between the wheels and the body of vehicles. The vibration energy of the vehicle body can be dissipated by the damping characteristics of the suspension system [1]. According to the adjustability of the damping force, the suspension system can be classified into three types: the passive suspension, the semiactive suspension (SAS), and the active suspension [2]. In these three types of suspension, the SAS system can achieve similar performance to the active suspension system while less energy consumption and low cost. Specifically, the magnetorheological fluid-based (MRF) SAS has been widely employed in vehicle suspension systems due to its faster response [3]. In the MRF-SAS system, the nonlinear hysteresis behavior of the MRF damper is pronounced and can be characterized by the hysteresis model [3]. Generally, the hysteresis model can be classified into two categories in terms of the phenomenological (PL) model and the black-box model [4]. The PL model can be classified into the B-W model [5], the sigmoid model [6], and the hyperbolic model [7]; and the black-box model can be classified into the fuzzy model [8] and the neural network model [9]. The black-box model can achieve high approximate accuracy by increasing the number of the internal connection points, whereas its computational efficiency is reduced. In contrast, the PL model [2], due to its simple mathematical expressions, is an effective method to trade off modeling accuracy and computational efficiency.
