Model Predictive Local Linear Controller with Adaptive Constraint Handling

This study presents a novel Model Predictive Local Linear Controller (MPLLC) designed to enhance the performance of automatic control systems. By optimizing a linear mapping from states to control inputs with constraint penalty mechanisms, we achieve precise prediction and control of system states, which better handles nonlinear systems. Unlike traditional MPC that optimizes control sequences, MPLLC optimizes a compact linear mapping, requiring shorter prediction horizons while maintaining control performance. Experimental results demonstrate that MPLLC performs excellently in handling complex dynamic systems, effectively reducing computational cost and improving system stability. This research provides new insights for further development in the field of automatic control.