Dual-Modal Filtered-x LMS Preview Control of an Active Suspension Using a Lotus Modal-Force Transformation

Abstract

This study proposes a dual-modal preview-control framework that combines two filtered-x least-mean-square (FxLMS) algorithms with a Lotus-type modal-force transformation for active suspension systems. Using previewed road information as a common reference, the heave- and pitch-mode FxLMS controllers independently generate a generalized vertical force and pitch moment to reduce sprung-mass vertical acceleration and pitch rate, respectively. These modal commands are mapped to the front and rear actuator forces through a full-rank modal-force transformation defined from the half-car geometry. Although the baseline and proposed architectures use the same two physical actuators, the proposed controller replaces the baseline's single adaptive rear-force correction with two independently adapted generalized commands, thereby separating the prescribed heave and pitch commands at the command-allocation level. Performance was assessed using conventional ride-comfort and motion-sickness dose measures together with supplementary visual-task-weighted indices. CarSim-MATLAB/Simulink co-simulations were conducted under four selected road-input conditions. Compared with the selected baseline, the proposed architecture produced lower vertical-motion indices in most cases, whereas the relative pitch-response benefit depended on the road input and evaluation metric. These results support further investigation under constrained and experimentally validated conditions.