Model-Free Approximated Optimal Trajectory Tracking Control of Photoelectric Tracking System: A Digital Twin Approach

The parameter and model uncertainties are inevitable in modern industrial systems. To address this problem, a digital-twin-based model-free approximated optimal trajectory tracking control strategy is proposed in this article for the photoelectric tracking system (PTS). With the online operational data, the digital replica of the physical PTS is dynamically established based on the designed parameter learning law to estimate the model parameters iteratively for the subsequent control optimization. To prevent the direct solving of Hamilton–Jacobian–Bellman differential equation, the performance index in this article is estimated by employing the Taylor series expansion, and a computationally efficient control law with exponential convergence is explicitly derived based on the projection neural controller for model-free approximated optimal control of PTS. Additionally, comprehensive numerical simulation and experiments on a laboratory prototype of PTS verify the excellent performance and practical feasibility of the proposed method even in scenarios of serious perturbations.