改进教与学优化算法的 LQR 控制器优化设计

摘要

为了快速有效地确定线性二次最优控制（ linear quadratic regulator， LQR）问题中的加权矩阵Q和R，针对主动悬架LQR控制器权系数设计问题，提出一种改进的教与学优化算法进行LQR优化设计。算法对基本教与学优化算法中的“教”与“学”阶段进行了进一步的改进，同时提出一种“自我学习”策略。通过仿真实验表明，和基本教与学算法、粒子群算法、遗传算法相比，本文算法在对主动悬架LQR控制器优化时，具有收敛速度快，求解精度高和稳定性强等优势。%To determine the weighting matrix Q and R for a linear quadratic regulator ( LQR) , a modified teaching-learning-based optimization ( MTLBO) algorithm is proposed to tune weighting factors for active suspension LQR controller.The “Teaching” phase and“learning” phase are modified using MTLBO based on the basic TLBO algo-rithm.A novel“self-learning” strategy is employed in MTLBO.The simulation results showed that the MTLBO algo-rithm has distinct advantages in convergence, precision and stability than basic TLBO, PSO and genetic algorithms.