Cooperative Swarm Cover Penetration Strategy Based Intercept Point Prediction by Neural Network

This paper addresses the challenge of hypersonic vehicle swarm penetration in adversarial environments by proposing a novel cooperative cover strategy that eliminates mid-phase interceptor information requirements. Firstly, The problem formulation and the penetration strategy is introduced. Secondly, a neural network-based intercept point prediction model leveraging observable parameters only during interceptor propulsion cutoff phases is established. After that, a consensus-driven guidance law to realize the strategy is designed. Finally, both single-time and Monte Carlo numerical simulations are conducted. The results demonstrate more than 71% successful penetration rates against large-scale '10 vs 10' interception scenarios, proving the effectiveness and robustness against random variations of missile initial conditions of the proposed strategy.