Data-Driven Fuzzy Adaptive Critic for Nonlinear Continuous-Time Systems Through Value Iteration

A novel model-free scheme combining value iteration (VI) with adaptive dynamic programming is developed to achieve optimal regulation of nonlinear continuous-time systems. By employing VI, the conventional need for admissible initial policies is circumvented while progressively approximating the optimal control solution. To avoid using the system model, an enhanced sparrow search algorithm is applied to modify traditional VI. An adaptive network-based fuzzy inference system is utilized to extract the required characteristics of the dynamical model. Finally, a benchmark case study is implemented to validate the efficacy of the established methodology.