Analysis Of The Hierarchical Target System Of Systems And The Generation Of Target Striking Actions

Modern Warfare has a significant feature of the System of Systems(So S).The important factor of the successes and failures of the warfare is striking the enemy So S efficiently to ensure the total failure of the enemy So S.The generation process of the target So S striking becomes a key factor in command decision.The hierarchical So S is a typical type of target So S.The generation of the hierarchical So S striking actions is affected by many factors,including the complex hierarchy-relation,the dynamic response characteristics,and the counter action of the enemy,etc.The traditional analytical models and optimization methods now in used have some defects such as weak modeling ability and low effectiveness of algorithms.In order to solve such problems,we divided the issue into three decision scenarios with typical significance militarily sense,and studied how to plan the target So S striking actions in these scenarios.The main contents of this paper can be summarized as the following four parts:(1)Established a formal description of the hierarchical target So S and the generating framework of So S strike actions.We formally described the elements and states of the hierarchical So S,including the elements,states and the So S failure mechanisms,which is the causality correlation of invalidation So S by strike actions.We analyzed the revolution process of striking-drived So S states,and then set up the models of target So S striking actions generation problem,and set up a basic framework of how to solve the problem.(2)Advance the generation method of hierarchical target So S Striking actions under the condition of no counterattack and single striking phase;Under the certain condition,the static fault tree method was introduced to describe the So S failure mechanism.The action generation model and a heuristic were given based on the mechanism,and the adaptive messy genetic algorithm was produced to generate the striking actions,which coded the decomposition choices of the static fault tree into the chromosome.Under the uncertain condition,The OOBN(Object Oriented Bayesians Network)method was introduced to describe the So S failure mechanism,which was composed by some bayes fragments of influential relations of different kinds in target So S.The action generation model and a heuristic were given based on the OOBN,and then a genetic algorithm with punishment functions was given to generate the striking actions under multiple constraints such as target selecting and resource limits.(3)Advance the generation method of hierarchical target So S Striking actions under the condition of no counterattack and multiple striking phase.Under the certain condition,the dynamic fault tree method is introduced to describe the So S failure mechanism,which established the hierarchical relations between the striking actions sequences and the failure event of target So S.The action generation model was given,and transformed the multiple phases striking decision problem into a problem which seek the optimum dynamic logical combination of bottom event of fault tree,and then a adaptive messy genetic algorithm which coded the action sequence information into the chromosome are given to solve the model.Under the uncertain condition,the Object Oriented Dynamic Bayesians Network(OODBN)method was introduced to model the So S failure mechanism,and a Markov Decision Process based attack model was provided to solve the target selecting policy.Based on the estimate value of the current state of target So S,a heuristic value was calculated to guide the target selecting policy search while reducing the search space of the Markov Decision Process.(4)Advance the generation method of hierarchical target So S Striking actions under the condition of counterattack and multiple striking phases.Under the certain condition,the game theory was introduced to model the So S striking process into multiple phase game,and transformed the problem into standard game tree which was solved by the backward induction method,and a method was given to simplify the action spaces of the game model.Under the uncertain condition,a target striking generation model based on the dynamic bayes net and the random game theory was given,and transformed the problem into standard game tree,and give the solving method.Finally,some cases were proposed to validate these methods.The results showed that it can solve the target striking actions policy under different decision scenarios.