Intelligent recognition method of target tactical behavior intention in air combat based on deep learning

Online and accurate target intention recognition has emerged as a promising research area for facilitating intelligent decision-making. It plays a crucial role in promoting significant advancements in situational awareness and creating tactical opportunities. To address the challenges posed by existing algorithms, such as their subjectivity, limited exploration of time-series characteristics, and failure to meet operational requirements, this paper presents a novel algorithm for target tactical intention recognition. We analyze the process of air combat to construct a model of target's tactical behavior. To the best of our knowledge, we present the first intuitive yet comprehensive analysis of tactical behavior, categorized into five distinct groups. Subsequently, we designed a tactical intention recognition method, which employs convolution, bidirectional long short-term memory (Bi-LSTM), and incorporates a self-attention mechanism. Convolution and Bi-LSTM were utilized to capture deep temporal features, and then were combined with the self-attention mechanism, effectively focusing on key features to eliminate interference, enhance the convergence speed, and improve recognition accuracy. The performance of the proposed method was evaluated in terms of reliability, accuracy, and timeliness by comparison with other algorithms. The results demonstrate the practical value of the method in air combat.