PreFuzzy-LADRC fusion control for magnetorheological semi-active suspension based on binocular vision preview and OBL-DE-PSO parameter optimization

Magnetorheological (MR) semi-active suspensions suffer from inherent challenges including damper nonlinear hysteresis, feedback control delay, and laborious manual parameter tuning. To address these issues, this paper proposes a preview fuzzy feedforward Lfal-based active disturbance rejection control (PreFuzzy-LADRC) strategy, where the Lfal-based active disturbance rejection control (LADRC) serves as the feedback core with a smoothed nonlinear function to suppress control chattering. An opposition-based learning and differential evolution enhanced particle swarm optimization (OBL-DE-PSO) algorithm is developed for offline tuning of the controller’s key feedback gains. The overall system integrates a Spencer forward damper model and an adaptive neuro-fuzzy inference system (ANFIS) inverse damper model, a two-degree-of-freedom (2-DOF) quarter-vehicle model, and a YOLOv8-enabled binocular vision system for road preview information acquisition. Simulations are conducted under ISO Class-B random roads and discrete impact roads. Compared with the empirically tuned PreFuzzy-LADRC, the OBL-DE-PSO-optimized controller reduces the root mean square (RMS) values of sprung mass vertical acceleration, suspension deflection, and tire deflection by 19.7%, 15.1%, and 22.5% under random road conditions, and by 9.3%, 19.3%, and 17.0% under discrete impact conditions, respectively. Hardware-in-the-loop (HIL) experiments validate the simulation results, with maximum relative deviations below 11%. Notably, the parameter optimization is performed offline, and the total online latency of visual perception and control computation satisfies the requirements of real-time applications. The results demonstrate that the proposed vision preview fusion framework and intelligent parameter tuning method can effectively improve the vibration suppression performance of MR semi-active suspensions.