Research on LQR control strategy of magnetorheological semi-active suspension based on improved sand cat swarm optimization algorithm

To address actuator saturation, nonlinear dynamics, and empirical parameter tuning in magnetorheological (MR) semi-active suspensions, this paper proposes an linear quadratic regulator (LQR) control strategy optimized by an improved sand cat swarm optimization (ISCSO) algorithm. Based on a quarter-car model, a feedforward neural network trained on experimental data is developed as a data-driven model of the MR damper. On this basis, an LQR controller is designed to map the ideal control force into the realizable force range under semi-active constraints. The ISCSO algorithm is further employed to optimize the LQR weighting matrices, with clipping limits incorporated into the fitness evaluation to reduce the mismatch between theoretical commands and physical constraints. Numerical simulations and real-vehicle road tests show that the proposed strategy outperforms passive suspension and standard sand cat swarm optimization-LQR in vibration suppression. In particular, under class c random-road excitation in simulation, the RMS body acceleration and suspension deflection are reduced by 35.13% and 39.86%, respectively. Road-test results further confirm its effectiveness in reducing body vibration while maintaining acceptable suspension travel performance. These results demonstrate that the proposed method provides an effective framework for constrained control of MR semi-active suspension systems.