Reinforcement Learning-Based Dynamic Programming Control for Robust Disturbance Rejection and Error Minimization in Dynamic Systems

Dynamic programming (DP) and reinforcement learning (RL) are effective techniques for maximizing control in uncertain and dynamic systems. In order to improve decision-making in control applications, this study combines RL and DP with an emphasis on error minimization and disturbance rejection. The strategy in the RL framework aims to maximize expected returns and is organized as a Markov Decision Process (MDP). The method allows for enhanced adaptive control techniques, error minimization, and disturbance rejection by integrating state and action value functions. In order to maximize system performance, the study also investigates dynamic control equations using the Bellman principle and Hamilton-Jacobi-Bellman formulations. The suggested approach improves overall system stability by increasing tracking speed, reducing control errors, and fortifying robustness against sudden disturbance.