El artículo propone una estrategia de defensa colaborativa para escenarios de persecución-evasión-defensa de naves espaciales. En estas situaciones, un perseguidor intenta capturar a un evasor mientras evita a un defensor, y el evasor y el defensor trabajan juntos para evitarlo. La estrategia se basa en la teoría de juegos diferenciales, identificando los modos del perseguidor (ataque y evasión) y rastreándolos en tiempo real utilizando un algoritmo interactivo de emparejamiento de modelos múltiples (IMM) con filtros de estructura variable suaves. Basándose en la identificación de estrategias y la teoría de optimización lineal-cuadrática, el evasor y el defensor adoptan una estrategia cooperativa, donde el evasor ayuda al defensor atrayendo al perseguidor para que lo intercepte. Las simulaciones muestran que el algoritmo IMM funciona bien al reconocer la estrategia del perseguidor, y la estrategia de defensa colaborativa intercepta eficazmente a los perseguidores que adaptan sus objetivos.
Abstract
This paper presents a mode-switching collaborative defense strategy for spacecraft pursuit-evasion-defense scenarios. In these scenarios, the pursuer tries to avoid the defender while capturing the evader, while the evader and defender form an alliance to prevent the pursuer from achieving its goal. First, the behavioral modes of the pursuer, including attack and avoidance modes, were established using differential game theory. These modes are then recognized by an interactive multiple model-matching algorithm (IMM), that uses several smooth variable structure filters to match the modes of the pursuer and update their probabilities in real time. Based on the linear-quadratic optimization theory, combined with the results of strategy identification, a two-way cooperative optimal strategy for the defender and evader is proposed, where the evader aids the defender to intercept the pursuer by performing luring maneuvers. Simulation results show that the interactive multi-model algorithm based on several smooth variable structure filters perform well in the strategy identification of the pursuer, and the cooperative defense strategy based on strategy identification has good interception performance when facing pursuers, who are able to flexibly adjust their game objectives.
Fuente: https://www.sciencedirect.com