Online Data-Driven-Based Optimal Output Tracking Control Without Initial Stabilizing Policy

This article investigates the optimal output tracking control problem for a continuous-time linear system with an unknown system model. By integrating adaptive dynamic programming with optimal control theory, a dual policy iteration (PI) learning algorithm composed of two PI schemes is proposed to adaptively learn the optimal tracking controller. The primary advantage of the proposed algorithm lies in that it does not require an initial stabilizing control policy, persistence of excitation, or storage of historical data to guarantee convergence. This feature fundamentally distinguishes it from existing approaches based on the least-squares method, which rely on these conditions. Simulation results demonstrate the effectiveness of the proposed algorithm, and its superiority is further validated through comparisons with existing methods.