Study On Double Linkage Simulaiton Optimization For Joint Fire Strike Mission Planning

The mean of joint strike fire is a combat action on important target to reach an object using many types of weapons.Joint Fire Strike,as an important combat style of joint campaign is attend more and more by the international community,because of involving many different military and high technology weapons,large amount of information and computation,timeliness,high accuracy requirement,the operation scheme optimization is complicated system engineering.Due to the complexity of the process of confrontation of joint fire strike system of systems target,there are many difficulties in solving such problem using case changing based,analytic model,modeling and simulation based methods.In view of this,the optimization problem is divided into two layers,the relation of the two layers is built,and different solutions are proposed according to different levels,in order to make full use of the information generated in the optimization process,to efficiently and credibly produce joint fire strike plan.Aiming at lack of domestic and foreign related technology in this research study,the main research contents and results are as follows:(1)Joint fire strike mission planning problem modeling and solution flamework are given.First the joint fire strike mission planning problem is modeled,and divided into task optimization and operation optimization,task optimization is target selection and resource allocation optimization,operation optimization is cooperative combat style optimization.Second,double linkage simulation framework is constructed tso combine the two optimization problems,both levels by means of supplying inputs for each to drive each carrying out optimization to fetch an best joint fire strike scheme.(2)Improved cognition evolutionary algorithm method for task optimization is given.In order to make full use of the optimization process information to improve the optimization efficiency,creative thinking is introduced into problem solving.First,based on fully express domain knowledge,easy realization and understanding of knowledge,easy maintenance and management of knowledge,easy assembly and reasoning of knowledge,the Bayesian network is introduced into knowledge representation.Second,in order to improve the learning accuracy,belief map and fuzzy set theory are used so as to more effectively acquire and fuse expert and process information,the MDL measure is proved,and the K2 algorithm is introduced into the structure learning.Third,the memory is studied from the perspective of information processing,and the method of information coding,information storage and information extraction are put forward.Fourth,thinking module is designed,the breadth and depth search optimization algorithm are instructed in divergent thinking so as to achieve the balance of diversity of solution and utilization of deep knowledge;in convergent thinking,based on second resource allocation,focused thinking and innovatethinking are combined to produce an attack architecture.(3)Adaptive simulation optimization method for operation optimization is given.In order to efficiently generate action plan for each combat mission,Active knowledge mining and optimization algorithm are introduced into problem solving.First,Based on the relationship between tasks,a method of simulation experiment space division is proposed,is also a method of problem division.Second,a new method for constructing optimal experimental design(SOEDM)is developed in this paper:(a)a multi-objective optimal criterion is developed by combining correlation and space-filling criteria,(a)an efficient global optimal search algorithm,named as improved enhanced stochastic evolutionary(IESE)algorithm is developed.Third,a meta-modeling method based on robust support vector machine is proposed,gradient method is used in the parameter optimization process based on problem characteristics to eliminate random factors inherent in the simulation data as possible.Fourth,a simulation runs control method based on improved confidence interval is proposed,to control the random factors inherent in the simulation data.Fifth,estimation of distribution algorithm is improved by combining elite strategy,crowding mechanism and truncation select to to achieve trade-offs optimal rate and optimal result.(4)The feasibility and validity of the proposed method is proven with a case study of submarine-launched and plane-launched anti-ship missiles against surface fleet.