Three-Dimensional Cooperative Guidance with Multiple Constraints Based on Proximal Policy Optimization

Collaborative guidance technology is a crucial means to enhance the effectiveness of strikes. To achieve precise collaboration among multiple missiles targeting a common objective, this paper addresses the issue of inaccurate calculation of the virtual impact point control expected flight time resulting from the use of fast iterative algorithms. We propose a method based on proximal policy optimization to calculate the virtual impact point. A collaborative guidance model under multiple constraints is established, and a proximal policy optimization algorithm is applied to optimize the collaborative guidance law. The calculation parameters of the virtual impact point are treated as actions of an intelligent agent acting on the environment, with velocity, desired pitch angle, and position coordinates serving as the algorithm's observations. A reward function reflecting the collaborative time is constructed, establishing a multi-constraint collaborative guidance law based on intelligent learning. Extensive simulation experiments targeting stationary targets demonstrate the rationality and effectiveness of the proposed method. After training, the intelligent agent provides different desired attack angles, and, based on the observation space, it can generate corresponding parameters. In some scenarios, the precision of hitting time surpasses that of fast iterative algorithms.