基于深度神经网络AdaMod优化模型的来袭目标攻击意图识别

Identification of Attacking Target"s Intention Based on the AdaMod Optimization Model of Deep Neural Network

摘要:

海上舰艇防空反导作战基于目标攻击意图识别是现代舰艇防空技术的研究热点。来袭目标攻击意图识别是战场态势分析的一个重要部分,以往是通过先验知识和先验概率进行量化分析与明确攻击意图识别特征值的影响权重。深度神经网络可通过自适应学习目标攻击意图的特征值,可以在缺乏先验知识的条件下,通过小样本集的目标攻击意图的特征值训练,学习特征数据和攻击意图识别之间的对应关系与映射。通过引入GeLUs(Gaussian Error Linear Units)激活函数和AdaMod优化算法加快模型收敛,并解决了Adam模型可能无法收敛到最优解的问题。实验结果显示文中提出的模型可以有效解决在先验知识不足及训练数据规模小的情况下,能够有效识别来袭目标攻击意图,同时保证更高的准确率。

Abstract:

Recognition of target attack intent in naval ship air defense and antimissile operations is a hot research topic in naval ship air defense technology. The attack intention identification of incoming targets is an important part of battlefield situation analysis. In the past, quantitative analysis was conducted through prior knowledge and probability, and the impact weights of attack intention identification feature values were determined. Deep neural networks can adaptively learn the eigenvalues of target attack intentions, and can learn the correspondence and mapping between feature data and attack intention recognition through eigenvalue training of target attack intentions in small sample sets without prior knowledge. By introducing the GeLUs activation function and the AdaMod optimization algorithm, the model convergence is accelerated, and the problem that the Adam model may not converge to the optimal solution is solved. Experimental results of the text show that the model proposed can effectively identify the attack intent of incoming targets with insufficient prior knowledge and small training data, while ensuring a higher accuracy rate.