Discrete‐Time Reinforcement Control for Constraints and Substantial Disturbances With Barrier‐Unity Function: DC Motor Experimental Validation

ABSTRACT This work introduces a novel model‐free output‐feedback control strategy designed for optimal regulation under stringent operational constraints. The proposed approach rigorously enforces strict limits on input, output, and tracking error signals. Its efficacy is unequivocally demonstrated through validation on a critical DC motor torque control system, showcasing effective and safe operation within predefined boundaries. A central innovation involves the deployment of a zero‐minimum barrier function for meticulous handling of both symmetric and asymmetric constraints, seamlessly integrated with an augmented utility function. The underlying control framework leverages fuzzy rule‐emulating actor‐critic networks to achieve highly adaptive performance without requiring an explicit dynamic system model. Online learning laws are formulated, which ensure near‐optimal tracking and guarantee the forward invariance of the safe operating region. Crucially, these laws operate without inner‐loop iterations, thereby significantly enhancing computational efficiency. The experimental results, obtained under severe external disturbances, unequivocally demonstrate the controller's superior robustness in preventing constraint violations. Comprehensive comparative analyses further underscore its distinct advantages in effectively balancing performance optimization with rigorous constraint enforcement, leading to exceptionally stable and reliable system operation.