The Trajectory Generation of UCAV Evading Missiles Based on Neural Networks

Abstract

Abstract In order to solve the problem of evading air-to-air missiles of UCAV in autonomous air combat, the flight dynamics model and 3-dimensional guidance trajectory model based on proportional guidance were established and performance constraint conditions of missiles were constructed. According to the definition of basic maneuver library, 72 kinds of avoidance maneuvers were constructed. All 72 avoidance maneuvers were simulated while UCAV was at different relative yaw angles and relative pitch angles, and the optimal avoidance maneuvers under the corresponding conditions were selected out. Training samples and test samples were constructed by utilizing the acquired data, and the neural network for generating control parameters was trained. Under different conditions, neural network method and random selection method were simulated. The results show that the success rate of escape of the proposed method is higher and the time cost of generating the control parameters is less, which meets the requirements of effectiveness and real-time.