IET Control Theory & Applications Special Section: Emerging Trends in LPV-Based Control of Intelligent Automotive Systems Model predictive control for integrated longitudinal and lateral stability of electric vehicles with in-wheel motorsISSN 1751-8644 Received on 29th January 2020 Revised 26th May 2020 Accepted on 29th June 2020 E-First on 9th September 2020 doi: 10.1049/iet-cta.2020.0122 www.ietdl.org Lin Zhang1,2, Hong Chen2,3 , Yanjun Huang2, Hongyan Guo4, Haobo Sun2, Haitao Ding3, Nian Wang5 1Postdoctoral Station of Mechanical Engineering, Tongji University, Shanghai 201804, People's Republic of China 2School of Automotive Studies, Tongji University, Shanghai 201804, People's Republic of China 3State Key Laboratory of Automotive Simulation and Control, Jilin University, Changchun, Jilin 130025, People's Republic of China 4Department of Control Science and Engineering, Jilin University, Changchun, Jilin 130025, People's Republic of China 5Dongfeng Motor Corporotion, Wuhan, Hubei 430058, People's Republic of China E-mail: chenhong2019@tongji.edu.cn Abstract: This study investigates an integrated wheel slip, yaw rate, and sideslip angle control via torque vectoring to improve both the longitudinal and lateral stability of electric vehicles (EVs) with four in-wheel motors. The algorithm is developed based on model predictive control (MPC) and thus can optimally reach a balance among different objectives while considering actuation and state constraints. Firstly, to deal with tyre non-linearity and variations in the lateral tyre forces due to changes in tyre slip ratios, the mechanism of using torque vectoring to improve vehicle stability is analysed. Then, a non-linear tyre model is introduced into the predictive model to characterise the tyre force coupling relationship. Here, a linear-parameter-varying (LPV) model is employed, which is derived by linearising the nonlinear vehicle model online. Moreover, the stability control of EVs with in-wheel motors is transformed into a constrained online optimisation problem and solved using the proposed LPV-MPC method. Finally, the proposed LPV-MPC is compared with some existing well-established techniques from literature in different test scenarios. The obtained results demonstrate that the LPV-MPC approach could reduce the computational burden and shows a precise longitudinal control and obviously improves the lateral stability. 1 Introduction Electric vehicle (EV) is a promising solution to the issues such as rising energy costs and strict regulation of emissions [ 1]. However, the safety, e.g. stability of themselves is always a problem. For example, when an EV is driving fast on a slippery road, the behaviours such as ‘drift’ and ‘sharp turns’ are more likely to happen, resulting in severe traffic problems [2]. EVs especially the ones with in-wheel motors have been attracting attention because of their exclusive features in enhancing vehicle stability control [ 3]. To improve the stability of such EVs, as the core of vehicle stability systems, the direct yaw moment control (DYC) is commonly used. Specifically, once the DYC system detects the unstable tendency, it will reduce the torque of the engine or independently apply the braking force to tyres. Nevertheless, the additional yaw moment generated by braking sacrifice the speed of the vehicle. Thus, it not only reduces the driving ability but also intervenes the driver. By contrast, a direct yaw moment can also be formed by torques applied on each wheels and it refers to torque vectoring control (TVC) [ 4]. TVC is usually used to improve the lateral stability without causing obvious deceleration. In addition, another challenging task of vehicle stability control is longitudinal control, i.e. wheel slip ratio control. However, due to the characteristics of tyres, the lateral force margin is severely reduced when the wheel slip ratio exceeding a certain threshold. At this time, either understeer or oversteer phenomenon could hap