Research Of Belief-Rule-based Reasoning Technology For Learning Air Combat Maneuvers

As the technologys of aviation weapon have been improved continuously,air combat decision depended on human being can’t meet the demands.The study of air combat maneuver-making based on Artifical Intergence has become popular.It’s valuable and significant to study air combat maneuver decision as the core of air combat maneuver-making.This thesis develops a new approach based on Belief-Rule-based system to make maneuver strategies in one-to-one air combat.The basic idea is derived from datas which come from training and actual combat.The belief rules are extracted from the datas.Then the system can be built based on the belief rules.In the system,the rule base is cut into sub-bases,and the parameters of the sub-bases are optimized by using the data.To ascertain the format of the belief rule,the aircraft is regarded as a point mass model,and the concrete aircraft model is introduced clearly in this paper.The innovations are pointed out as follows.1.For solving the problem of uncertainty in combat data,the belief rules are constructed.The procedure contains three parts,which are discretization of the data,classification of the data and extraction of the belief rules.In the process of belief rule extraction,rough set support is used to represent the degree of belief.2.The factor of belief-rule reasoning speed and precision are analyzed,then,an approach to segment belief rules and optimize the parameters of the sub-base system is presented,which ensures the reasoning speed and accuracy rating.3.For resolving the problem of large number of sub-base system and the low efficiency of traditional parameters optimization methods,an improved genetic algorithm based on motivating factor is developed to optimize the parameters of the sub-base system.The motivating factor is forecasted by the GM(1,1)model based on the ranked object values.The operator is suitably used to improve the genetic algorithm.The improvements adjust the population continuously and give each individual more reasonable rate of crossover and mutation.4.Because the decision results based on belief rules have discrete distribution,this paper conducts second fusion to make the result continuous,which improves the system’s usability and practicability.At the meantime,the model of two airplane’s combat based on influence diagram is built to collect combat data.And the data is used to build air combat maneuver decision system based on belief rules.The system’s validity and rationality is tested through simulation experiments which also proves that the standard building process of decision system based on data can learn maneuver strategy from air combat data,and the approach is practicable in the aspect of one-to-one air combat.