Unmanned chariot target search strategy generation method

The invention relates to an unmanned combat vehicle target search strategy generation method, which comprises the steps of constructing an unmanned combat vehicle target search strategy generation model which comprises a state set of a plurality of unmanned combat vehicles at any moment and an action set of the plurality of unmanned combat vehicles at any moment, the environment feedback is obtained by the multiple unmanned combat vehicles; training the unmanned combat vehicle target search strategy generation model by using a meta-reinforcement learning algorithm; in a training process of introducing the weighted loss function, obtaining a target search strategy of the plurality of unmanned combat vehicles; and deploying the obtained target search strategies of the plurality of unmanned combat vehicles in a real task environment. According to the method, the problem of strategy model partial estimation in a meta reinforcement learning algorithm can be solved, and the efficiency and the quality of unmanned co