Safety-constrained transient control for aero-engines: A data-driven diffeomorphic ADP framework.

To enable safety-constrained control of aero-engines under wide-range transient conditions, a novel data-driven diffeomorphic adaptive dynamic programming (ADP) framework is developed to explicitly enforce the state and input safety constraints. The approach begins by employing diffeomorphic transformations coupled with a dynamic control law to effectively eliminate explicit state constraints. This transformation reformulates the original constrained problem into an optimal control framework subject solely to virtual input saturation. Subsequently, to handle input constraints, an inverse hyperbolic tangent barrier function is designed, thereby facilitating the application of Bellman's optimality principle. Leveraging this framework, a data-driven policy iteration method is developed to efficiently approximate the solution of the Hamilton-Jacobi-Bellman equation and derive the optimal control law. Rigorous theoretical analysis confirms the feasibility and stability of the proposed approach. Extensive simulations on a JT9D engine validate the proposed method's capability in achieving safe and rapid operating condition transitions. Compared to traditional PID control and particle swarm optimization based model predictive control, the proposed method achieves superior control performance, reduces the acceleration time by 24.6 %, and significantly lowers computational complexity. This research presents a promising and practical solution for advanced aero-engine control.