Abstract: Based on the state feature of the target, traditional methods of inferring aerial targets’ combat intention, require a large amount of domain expert knowledge to quantify the weight of feature, prior probability and clear correspondence between state characteristics and intention. However, the neural network method can train itself to obtain the rules between state features and intentions in the absence of domain expert knowledge. The Back Propagation (BP) algorithm has slow convergence rate and can easily fall into local optimum when updating the network node weights. Through the introduction of the Rectified Linear Unit (ReLU) activation function and the Adaptive moment estimation (Adam) optimization algorithm, a combat intention recognition model based on deep neural network is designed to improve the convergence speed of the model and effectively prevent the algorithm from falling into a local optimum. The simulation results show that the proposed method can effectively recognise the aerial targets’ combat intention and obtain a higher recognition rate.

Key words: combat intention, expert knowledge, neural network, recognition, aerial target