MADDPG-based unmanned aerial vehicle group air interception task allocation method

The invention discloses an unmanned aerial vehicle group air interception task allocation method based on MADDPG, and relates to the technical field of multi-agent collaborative decision making, and the method comprises the steps: firstly, carrying out the multi-agent reinforcement learning core element design for an unmanned aerial vehicle group air interception scene; secondly, an invalid action mask is introduced on the basis of an MADDPG algorithm, invalid exploration of an intelligent agent is avoided based on human knowledge, the model training efficiency is improved, and therefore autonomous collaborative distribution of unmanned aerial vehicle group detection/tracking/strike tasks is achieved; the invention aims to establish a detection/tracking/interception task autonomous allocation model for an air interception scene of an unmanned aerial vehicle group, introduces an invalid action mask on the basis of an MADDPG algorithm, and realizes task autonomous coordination allocation between unmanned aerial