Study On Intelligent Decision-Making Of Air-to-Ground Attack With UCAV Teams Based On Bayesian Networks

The research of decision-making of the UCAV teams air-to-ground attack(UCAVTAGA) covers and integrates many scientific fields. Therefore, no systemicresearch exists in our country at present and such research in other countries is also inthe experimental validation phase. The decision-making system of the UCAVTAGA isthe central system of the UCAV teams. How to make an effective inference according tothe uncertain evidence acquired is a problem needed an urgent solvent. This paper issupported by the fund of Shanxi Natural Science. Aiming at realizing a quick andcorrect intelligent decision-making of the UCAVTAGA, this paper studied the mainproblems of UCAVTAGA based on the powerful inference capability of the Bayesiannetwork (BNs) and the advantage of the graph in expression. The research is dividedinto the following 7 parts:1) Study on the Intelligent Decision-making System of UCAVTAGAThe system's structure and functional classification of UCAVTAGA have beenfirstly presented based on the consistence of the attack-defense confront system. Thecombat flow and attack mode of UCAVTAGA have been put forward. Interactivestrategy of information, knowledge type required, knowledge achieve mode andknowledge expression of the UCAVTAGA has been put forward toward the complexityand distribution of knowledge. The intellectual decision-making system of theUCAVTAGA and the key technology required in the decision-making system has beenalso proposed.2) Study on the Performance of Probability Distribution Estimation Based onDecision Graph and Bayesian NetworksDecision-Graph Bayesian Optimization Algorithm (DBOA) is a probabilitydistribution estimation algorithm. The flow and primary operation of the algorithm havebeen analyses in detail. Moreover, the effect on the performance of the algorithm causedby the key parameters has been conducted a quantitive research. The superiority insolving the decomposable problems has been validated with comparing with the geneticalgorithm, PSO and Bayesian optimization algorithm.3) Study on Distributed Situation Assessment of UCAVTAGA Based on BNThe space expression and mathematical mode of the situation assessment of theUCAVTAGA have been researched. Due to the BN is adapted to the multi-sensor fusion,the distribution situation assessment of the UCAVTAGA using Bayesian network hasbeen brought forward. Tactical situation assessment mode of UCAVTAGA based on BNhas been built. Multi-source information fusion inference algorithm and sensitiveanalysis method using Bayesian network have been conducted and simulated.4) Study on the Cooperate Task Assignment of UCAVTAGA Based on DBOA Task assignment algorithm of UCAVTAGA based on DBOA is presented. Theeffect caused by electronic confront has been taken into account. The concept ofDistance Discount Factor (DDF) has been introduced to address the fact that targetingclose but less significant units could be more rewarding than targeting far but moresignificant units. Simulation results have verified that the method could be used to solvethe complex question, the operation was quickly and the solution was best and the DDFis validity with comparing the simulation results with and without DDF.5) Study on the Tactical Task Decision-making of UCAVTAGA Based onInfluence Graph (IG)Tactical task decision-making method of the UCAVTAGA based on IG has beenbrought out. IG expands the dynamic Bayesian network with adding the decision nodeand value node to the Bayesian network. Because the decision-making of theUCAVTAGA is full of uncertainty, assistant decision-making in order to solve theproblem of dynamic tactical task decision-making is put forward based on IG mode andcombined with the value theory. The system structure, mathematical mode and relevantIG mode of the tactical task decision-making of UCAVTAGA have been built and thesimulation analysis has been conducted.6) Study on the Battle Damage Assessment of UCAVTAGA Based Fuzzy BNBattle damage assessment (BDA) algorithm based on the fuzzy Bayesian networkis presented for the first time. Fuzzy Bayesian network integrates the advantages of thefuzzy logic and BN, converting the clear node variable to fuzzy node variable. Thispaper also presents a method to solve the BDA based on BN. BN mode of BDA hasbeen built and simulated.This paper depicts the decision-making mode of the UCAVTAGA in the futurebattlefield through the analysis of intellectual decision-making system. Moreover, thekey technology is simulated and modeled based on BN and the effectiveness of themode and algorithm is proved according to the simulation, which providing a solidfoundation for the analysis and improvement of the automation and intelligence of thedecision-making process of UCAVTAGA.