Research Of UCAV Air Combat Based On Reinforcemnt Learning

With unmanned combat air vehicles in the military field is applied more and more widely, the research is developing in the world. UCAV will completely change the future air combat situation, has become an important force in the battlefield. The intelligent level of UCAV will be the key to the development of UCAV Technology. The UCAV will not only be able to load information through the sensor s, but also participate in air combat. The air combat decision-making theory has been developed in recent decades.It has gradually formed two methods, one is expert system, differential game, etc.,and the other is neural network, genetic algorithm and so on, etc.In this paper, we will use reinforcement learning to solve the air combat decision-making,the main work is as follows:First, this paper introduces the basic principle of reinforcement learning algorithm, studies on the typical reinforcement learning algorithm,introduces the application field of reinforcement learning.Scend, based on the research hotspot and the classical of reinforcement learning, improve algorithm performance.This chapter introduces the motivation layer, the two layer structure is changed into three layers structure, the prior knowledge introduces to the reinforcement learning algorithm, improved the speed of algorithm. Then, the idea of simulated annealing algorithm is introduced, the performance of the algorithm is proved by experiments.Third, focus on the air combat decision, on the basis of neural network,the motivation guided reinforcement learning algorithm is designed.The effectiveness of the algorithm is verified by comparing the flight trajectory of different decision intention.