Journal of Mechanical Science and Technology 30 (8) (2016) 3835~3845 www.springerlink.com/content/1738-494x(Print)/1976-3824(Online) DOI 10.1007/s12206-016-0747-8 Comparative study of MPC and LQC with disturbance rejection control for heavy vehicle rollover prevention in an inclement environment † Fitri Yakub 1,2,* Shihao Lee 3 and Yasuchika Mori 2 1 Malaysia-Japan International Institute of Technology, Universiti Teknologi Malaysia, Jln Semarak, 54100, Kuala Lumpur, Malaysia 2 Graduate School of System Design, Tokyo Metropolitan University, Hino, 191-0065, Tokyo, Japan 3 Hitachi LTD, Hitachi, 317-8601, Ibaraki, Japan (Manuscript Received February 8, 2015; Revised July 5, 2015; Accepted April 11, 2016) ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Abstract This paper compares the Model predictive control (MPC) and Linear quadratic control (LQC) of heavy vehicles via active front steer- ing for rollover prevention in inclement environments. In both control methods, a Disturbance rejection control (DRC) that negates the effects of wind and road bank acting on the vehicle is designed. Load transfer ratio (LTR) is applied to judge rollover by mitigating the absolute value of LTR as much as possible. We tested and compared two different controllers, i) MPC with DRC and ii) LQC with DRC. Two types of environmental conditions were considered, i) typhoon and ii) typhoon on a bumpy road. The simulation results demonstrate that MPC was more successfully implemented than LQC during LTR suppression. This paper also proposes an MPC for coordination of active rear steering and differential braking control maneuvers to prevent rollover in inclement environments. For a feasible comparison, the LQC controller was designed using the same approach adopted for the MPC controller. Results show the proposed coordinated con- trol yields better performance for rollover prevention than LQC. Keywords: Heavy vehicle; Model predictive control; Linear quadratic control; Rollover prevention; Disturbance observer; Disturbance rejection control ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- 1. Introduction Inclement environmental conditions are an important factor in fatal rollovers. A strong lateral wind or significant road bank may harass heavy vehicle behavior, and even initiate rollover accidents, because roll stability is easily affected by these disturbances. Developing a control system for distur- bance detection and rollover prevention is important for vehi- cles moving on an uneven surface or through strong lateral winds. The disturbance observer (DOB) is an effective compensa- tion mechanism that reduces the effects of disturbances, un- certainties, and nonlinearities within the plant, and enforces nominal input/output behavior, particularly in the low fre- quency range, where the reference signal frequency is concen- trated. Accordingly, the DOB has the ability to reject high- order and stepwise disturbances asymptotically. In the past few years, the force of wind and moment act- ing on a vehicle body have been regarded as an unmeasured disturbance that can be estimated and suppressed without changing the input-output behavior by a DOB [1]. Its appli- cation to compensate for electric power steering is used to revise yaw rate performance [2]. In Ref. [3], low-order DOB was proven to successfully reduce the computational cost of implementation. The road bank angle can also be precisely estimated using a DOB from a Global positioning system (GPS) and inertial navigation system [4], or from low-cost onboard sensors [5]. Moreover, Ref. [6] presented road bank estimation using a dynamic simplex method for rollover prediction. Model predictive control (MPC) and Linear quadratic con- trol (LQC), which are based on a quadratic cost function, are widely implemented for process control [7], motion control [8] and vehicle control [9]. The main advantage of LQC is that optimal input signal can be obtained from full state feedback, whereas MPC is optimal when implemented in closed-loop systems. However, LQC has limitations in sys- tems affected by actuator limitations. For instance, restrict- ing the manipulated variable or the controlled variable can be difficult [10]. However, because by design, MPC does not directly handle disturbances, it cannot satisfactorily achieve control vehicle stability in the presence of strong disturbances and large un- certainties. Therefore, DOB with Disturbance rejection control (DRC) acts as an observer and compensator, thereby improv- ing vehicle performance. DOB-MPC-based approaches for * Corresponding author. Tel.: +60 11 5221 4962, Fax.: +60 3 2203 1266 E-mail address: mfitri.kl@utm.my † Recommended by Associate Editor Deok Jin Lee © KSME & Springer 2016