Three dimensional coupled sliding mode guidance law based on reinforcement learning

Aiming at the problem of parameter affecting guidance accuracy in the design of three-dimensional coupled sliding mode guidance law considering terminal angle constraints in the real battlefield. In this paper, a novel three-dimensional coupled intelligent algorithm for missile intercept maneuvering target is proposed by combining the reinforcement learning algorithm with the traditional sliding mode variable structure control. In this algorithm, the terminal guidance problem of missile intercept maneuvering target is regarded as a Markov decision-making process, and the Actor and Critic neural networks in the TD3 algorithm are utilized, at the same time, the integer reward function is designed according to the task requirements, which constantly interacts with the environment to update the network parameters and obtain the guidance law in the complex environment. Simulation experiments show that the proposed intelligent algorithm has better robustness and can realize the optimal selection of parameters in the designed sliding mode guidance law, and the proposed algorithm has better autonomous decision-making ability compared with the traditional sliding mode variable structure guidance, which can simultaneously optimize the ballistic curve in accomplishing the goal of minimal miss distance interception of maneuvering targets.