Intelligent Decision Making Of Multi-UAV Collaborative Attack And Defense Based On Intention Prediction

In recent years,as an emerging force in modern warfare,UAV have played an increasingly significant role in air combat,and have received widespread attention from major military powers.As part of UAVs technology,UAVs air combat autonomous decision-making is a key technology to enhance UAVs autonomy,which has aroused great interest from many scholars at home and abroad.However,with the continuous development of military science and technology,the air combat environment is becoming increasingly complex.To occupy air combat initiative in a complex environment,we must accurately understand the tactical intentions of enemy targets.In the background of UAV air combat in complex environments,this paper studies the multi-aircraft cooperative offensive and defensive decision-making for prediction of UAV air combat intentions.Firstly,this paper describes the target intention prediction,target assignment of multiple UAVs and air combat maneuver decision-making in detail,and establishes the overall structure of UAV intelligent attack and defense decision-making system.Then,aiming at the problem prediction of UAV air combat targets under incomplete information,an LSTM-based air combat target intent prediction model based on incomplete information is proposed.A hierarchical method is used to establish the air combat target intention prediction feature set,and the air combat information Coding into time series features,encapsulating expert experience into labels,introducing cubic sample interpolation function fitting and average filling method to patch incomplete data,using Adam optimization algorithm to speed up the training speed of target intent prediction model in order to effectively prevent local optimization The accuracy and validity of the intent prediction model and solving method are verified by simulation.Next,for the multi-UAVs cooperative target assignment problem,a multi-aircraft cooperative target assignment algorithm based on an improved discrete wolf pack algorithm is designed.A multi-aircraft cooperative target assignment model is established by comprehensively considering the air combat situation,fighter performance,and target intent.Considering,the single-to-single,single-to-multiple,and multi-to-multiple allocations in UAVs air combat,this paper designs a virtual sequence encoding method,and designs the intelligent behavior of the discrete wolf pack algorithm,in addition,introducing the head wolf walking mechanism to strengthen the head wolf search ability.The adaptive running stride method is adopted to accelerate the convergence speed of the discrete wolf pack algorithm,and the effectiveness of the multi-machine cooperative target assignment algorithm based on the improved discrete wolf pack algorithm proposed in this paper is verified by simulation experiments.Finally,aiming at the decision-making problem of UAV air combat maneuver,the air combat situation model is established according to the attack effect of air-to-air missile on different areas,and the UAV maneuver decision-making model is constructed by combining the UAV kinematics model and the basic maneuvering database.In addition,the probabilistic neural network is used to predict the enemy target maneuver according to the enemy target aircraft considering the air combat situation and the target intention The method of reinforcement learning is used to select our maneuver strategy,and the rationality and effectiveness of the maneuver prediction method are verified by simulation experiments.