Computational Intelligent Guidance Law Accounting for Impact-Angle-Constrained Based on Online Integral Reinforcement Learning

This paper presents a computational intelligent guidance law that allows missile to intercept target with predetermined impact angle. This problem is solved in the framework of nonlinear optimal control theory, the solution of which relies on the solution of Hamilton-Jacobi-Bellman equation. Because of the Hamilton-Jacobi-Bellman equation is a nonlinear partial differential equation and is intractable to find the analytical solution, an off-policy integral-reinforcement-learning algorithm which is a machine learning algorithm is used to solve the approximate optimal solution of the Hamilton-Jacobi-Bellman equation. The off-policy learning method is a model-free method, in which none of the knowledge of the dynamic system is required. Two neural-networks are used to implement the off-policy integral-reinforcement-learning algorithm, one is critic neural-network and the other is actor neural-network. By online data sampling, the approximate optimal solution of missile target interception problem can be learned online. Numerical simulation is conducted to validate the effectiveness of the computational intelligent guidance algorithm.