A new neuro-based solution for closed-loop optimal guidance with terminal constraints

A general optimal feedback controller can be obtained by solving the Hamilton Jacobi Bellman dynamic programming equations. But for nonlinear systems, such as homing missiles, it is a tedious task. In this paper, a new neuro-based solution has been used to design a feedback or closed-loop optimal guidance law. The methodology uses the Hamiltonian equations associated with optimal control theory and adapts values of the synaptic weights until closed-loop optimal control is achieved. Flight paths (trajectories) and the performance of the neurocontroller are compared to that of an exact optimal solution. Results show that the proposed method yields optimal closed loop control and can be used as an effective real time guidance method for nonlinear dynamical systems.