Data-Driven-Based Optimal Guidance Without Maneuverability Advantage

This paper investigates the energy-optimal guidance for attacking an target with equal maneuverability. According to the maximum principle, the optimality conditions for optimal interception problem under the acceleration ratio constraint are first established. Secondly, based on the optimality conditions, a parameterized system is designed that shares the same solution space as the original problem. This transforms the challenging nonlinear two-point boundary value problem into an equivalent integration problem. Then, the parameterized system is simply propagated to generate enough sampled data, which encapsulates the mapping relationship between the states and the optimal guidance commands. Finally, based on the sample data set, the neural network is trained to fit the most useful control. The superiority of the proposed method was demonstrated through comparative simulations.