Intention recognition of aerial targets based on Bayesian optimization algorithm

A Bayesian network optimization algorithm is proposed for the problem of combat intention identification of aerial group targets. The external characteristics of the data link of the targets group are extracted as the network nodes. Then the initial Bayesian network model is established with the network nodes and conditional probability tables. Simulating to get the sample data set, and evaluation of the network is made with the BIC measure. With this data set, the original prediction model is optimized using the method of Bayesian optimization algorithm. The new model's BIC score is lower than the original one which means the former is more matching with the dataset than the latter. Multiple sample sets of different quantities are tested and every test obtains a lower BIC score compared to the original model. The effectiveness of the optimization method is indicated.