Scholarlink

<p>Advanced methodologies for implementing adaptive-optimal control systems in multivariable environments are presented in this study, focusing on tank-level systems, where proportional-integral-derivative (PID) controllers are dynamically tuned via heuristic dynamic programming (HDP) and action-dependent heuristic dynamic programming (ADHDP) techniques. These adaptive strategies are compared to the traditional Schur method used to solve the Hamilton-Jacobi-Bellman (HJB) equation. The proposed architecture incorporates a recursive online tuning mechanism based on cost function approximation and reinforcement signals, and it is based on a complete system model and evaluated using a discrete linear quadratic regulator (DLQR) in a simulated water supply plant. Quantitative metrics and convergence properties are analyzed to validate the superiority of HDP and ADHDP over static gain approaches. This study contributes to the field of dynamic system control, highlighting the application of adaptive techniques in complex systems and in water resource management.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Optimal Online DLQR Design in MIMO Systems With HDP and ADHDP Approaches in Controlling Liquid Levels in Tanks

  • 期刊论文

  • 发表日期：

    2025年12月5日

查看全文
  • Evandro Martins Araújo Filho,
  • João Viana da Fonseca Neto,
  • José Pinheiro de Moura

摘要

Advanced methodologies for implementing adaptive-optimal control systems in multivariable environments are presented in this study, focusing on tank-level systems, where proportional-integral-derivative (PID) controllers are dynamically tuned via heuristic dynamic programming (HDP) and action-dependent heuristic dynamic programming (ADHDP) techniques. These adaptive strategies are compared to the traditional Schur method used to solve the Hamilton-Jacobi-Bellman (HJB) equation. The proposed architecture incorporates a recursive online tuning mechanism based on cost function approximation and reinforcement signals, and it is based on a complete system model and evaluated using a discrete linear quadratic regulator (DLQR) in a simulated water supply plant. Quantitative metrics and convergence properties are analyzed to validate the superiority of HDP and ADHDP over static gain approaches. This study contributes to the field of dynamic system control, highlighting the application of adaptive techniques in complex systems and in water resource management.