半主动悬架的舒适与安全协调控制策略是悬架控制的一个难题，如磁流变阻尼器有着较小的零场阻尼，在带来舒适性和大的阻尼力可调范围的同时也让轮胎更容易跳离地面，使得操控性乃至安全性出现问题，因此设计一种能够保证安全性的控制算法是必要的。预瞄控制则需要鲁棒的路面观测算法，能够充分利用传感器信号，为控制器提供可靠的路面信息。本文以模型预测控制为主要框架，对磁流变半主动悬架控制算法的设计进行了比较全面的对比研究，为基于悬架动力学的路面观测和磁流变悬架的模型预测控制设计提供良好的参考和依据。主要的工作内容和结论如下：1）建立了基于YK参数化方法的鲁棒路面观测器。通过簧下加速度传感器信号以及悬架变形速度对路面进行时域观测，从不同路面状况以及参数变动时的仿真结果来看，该方法相比较于传统的卡尔曼观测器具有更小的误差和更强的鲁棒性，能为车辆的预瞄控制提供更加准确的信息；2）设计并对比了磁流变悬架的线性MPC、混合MPC和非线性MPC控制算法。通过基于真实数据的模型在环仿真全面的研究并对比了3种设计方案，为后续的应用提供了参考与依据。仿真结果表明三种控制方案均能有效增强悬架的综合性能：其中线性MPC虽然无法考虑耗散性约束但是仍然是十分可用的解决方案；非线性MPC能够有效的对模型进行描述，实现更加精细的控制，具有很大的使用前景；而混合MPC更加适用于切换控制系统。此外，仿真结果还显示预瞄的加入与预测步长的增加均能显著的增强中低频下的控制效果，控制器代价函数的参数也会显著的影响控制效果；3）搭建了MPC控制算法的磁流变悬架试验台架。在不同工况下试验验证了MPC算法能够在保证车辆附着的同时提高舒适性，证明了算法的可用性以及MPC控制下的磁流变悬架能够实现操控性、舒适性和安全性的有机结合。其中，随机路面下的结果表明MPC算法能够在全频段实现折衷的性能，扫频路面试验结果进一步证明MPC算法能有效抑制轮胎离地的情况同时提高舒适性，冲击路面结果证明了MPC算法可以较为快速的衰减冲击带来的影响。

The comfort and safety coordination control strategy of semi-active suspension is a difficult problem in suspension control. For example, the magneto-rheological damper has a small zero-field damping ratio, which not only brings comfort and a large adjustable range of damping force, but also makes tires are more likely to jump off the ground, causing problems in handling and even safety. Therefore, it is necessary to design a control algorithm that can ensure safety. The suspension preview control requires a robust road profile observation algorithm, which can make full use of sensor signals and provide reliable road surface information for the controller. Based on the model predictive control as the main framework, this paper conducts a relatively comprehensive comparative study on the design of the magneto-rheological semi-active suspension control algorithm and provides a good reference and basis for the road surface observation based on suspension dynamics and the model predictive control design of magnetorheological semi-active suspension. The main work contents and conclusions are as follows:1) Establish a robust road profile observer based on the YK parameterization method. The time-domain observation of the road profile is carried out through the signal of the unsprung acceleration sensor and the deflection speed of the suspension. From the simulation results of different road conditions and parameter changes, this method has smaller errors and stronger robustness than the traditional Kalman observer, can provide more accurate information for the preview control of the vehicle; 2) Design and compare the control algorithms of linear MPC, hybrid MPC and nonlinear MPC of magnetorheological suspension. Through the comprehensive research of model-in-the-loop simulation based on real data and the comparison of three design schemes, it provides a reference and basis for subsequent applications. The simulation results show that all three control schemes can effectively enhance the comprehensive performance of the suspension: although the linear MPC cannot consider the dissipation constraints, it is still a very usable solution; the nonlinear MPC can effectively describe the model and achieve a more refined control, has great application prospects; and the hybrid MPC is more suitable for switching control systems. In addition, the simulation results also show that the addition of the preview and the increase of the prediction step can significantly enhance the control effect at low and medium frequencies, and the parameters of the controller cost function will also significantly affect the controller performance;3)Build a magnetorheological suspension test bench with MPC control algorithm. Experiments under different working conditions have verified that the MPC algorithm can improve the comfort while ensuring the vehicle adhesion, and prove the usability of the algorithm and the magneto-rheological suspension under MPC control can realize the organic combination of handling, comfort and safety. Among them, the results on random roads show that the MPC algorithm can achieve compromise performance in the whole frequency band. The results of the sweeping road test further prove that the MPC algorithm can effectively suppress the tires from the ground while improving comfort. The results of the impact on the road surface prove that the MPC algorithm can quickly attenuate the impact of the shocks.