Investigation into Airborne Target Intent Recognition Technology Utilizing Extreme Learning Machine
Zhige Xie, Chenguang Yang, Aizhi Liu, Lei Zhao · 2024
The determination of aerial target intent is a pivotal element for effectual aerial countermeasure training. Conventional rule-based strategies depend on an extensive assortment of domain-specific knowledge, whereas recognized target recognition techniques including template matching, knowledge mapping, Bayesian networks, neural networks, decision trees, and Hidden Markov Chains display notable limitations such as protracted training periods, intricate parameter optimization, and inconsistent recognition accuracy. To mitigate these limitations, this paper introduces a novel approach to aerial target intent recognition by deploying a extreme learning machine. The distinctive advantage of this method is its ability to leverage the rapid training speed and robust generalization capability of the superior learning machine, thereby enhancing the speed and precision of airborne target identification. Empirical findings from simulation experiments, complemented by subsequent comparative analysis with alternative methods, suggest that the proposed approach offers a balanced trade-off between training velocity and recognition accuracy. Therefore, it represents a promising progression in the realm of airborne target intent recognition.