基于混合注意力CNN的角反射器阵列部署

Deployment of Corner Reflector Arrays Based on Hybrid Attention Convolutional Neural Network

  • 摘要: 【目的】针对海面角反射器阵列排布问题,提出一种基于混合注意力CNN的优化角反射器阵列的部署方法。以提升角反射器对复杂目标的干扰能力,增强海上作战平台的电磁防护效能。【方法】首先,采用弹跳射线法对不同排布下的角反射器阵列几何模型进行电磁散射特性求解,得到一维距离像;其次,利用CLEAN算法从一维距离像中提取散射中心,构建包含散射中心与雷达入射角数据的数据集;然后,搭建融合混合注意力机制CNN的角反阵列排布预测网络模型,将数据集输入模型进行训练,以实现对角反射器排布的智能预测;最后,测试网络模型对输入不同的舰船一维距离像的预测能力,并将原始舰船与预测阵列排布的一维距离像进行对比分析。【结果】结果显示,该网络的训练损失为0.00027,且准确率相比原始CNN网络提升5.52%,表明模型具有较高的预测精度。在舰船目标简单散射特征场景下,由该方法布设的角反射器阵列一维距离像与舰船呈现高相关性,皮尔逊相关系数达到0.9137,在少数复杂情况下,该方法仍能借助耦合散射中心实现干扰规划。【结论】研究证明了基于混合注意力CNN的角反射器阵列部署方法能够有效优化角反射器的排布,提升其对目标的干扰能力,并为海上作战平台的电磁防护提供了新的技术途径。

     

    Abstract: Objectives This study proposes an optimization method for the deployment of corner reflector arrays based on a hybrid attention CNN(Convolutional Neural Network), aiming to address the arrangement challenges of sea surface corner reflector arrays. The objective is to enhance the interference capability of corner reflectors against complex targets and improve the electromagnetic protection efficacy of maritime combat platforms. Methods Initially, the electromagnetic scattering characteristics of corner reflector array geometric models under various arrangements were analyzed using the SBR (shooting and bouncing ray method), yielding one-dimensional range profiles. Subsequently, the CLEAN algorithm is applied to extract scattering centers from these one-dimensional range profiles, creating a dataset that includes data on scattering centers and radar incidence angles. Then, a prediction network model for corner reflector array arrangement is built by integrating a CNN with a hybrid attention mechanism. The dataset is input into the model for training to achieve intelligent prediction of corner reflector arrangement. Finally, the model's predictive performance is tested with different ship range profiles, and its predicted corner reflector arrays are compared with actual profiles to evaluate accuracy. Results The results show that the training loss of this network is 0.00027, and the accuracy rate is increased by 5.52% compared with the original CNN network, indicating that the model has high prediction accuracy. In scenarios with simple scattering features of ship targets, the one-dimensional range profile of the corner reflector array arranged by this method has a high correlation with that of ships, with the Pearson correlation coefficient reaching 0.9137. Even in a few complex situations, this method can still achieve interference planning by leveraging coupled scattering centers. Conclusions The research validates that the corner reflector array deployment method based on hybrid attention CNN effectively optimizes the arrangement of corner reflectors, enhances their interference capability against targets, and provides a novel technical approach for the electromagnetic protection of maritime combat platforms.

     

/

返回文章
返回