多元声阵的低小慢目标感知AI算法  

Research on Low and Slow Target Perception AI Algorithm based on Multiple Array

作　　者：苏晨 国蓉[2] 唐舒 袁成 SU Chen;GUO Rong;TANG Shu;YUAN Cheng(School of Electronic Information Engineering,Xi’an Technological University,Xi’an 710021,China;School of Opto electronical Engineering,Xi’an Technological University,Xi’an 710021,China;School of Defence Science and Technology,Xi’an Technological University,Xi’an 710021,China)

机构地区：[1]西安工业大学电子信息工程学院,西安710021 [2]西安工业大学光电工程学院,西安710021 [3]西安工业大学兵器科学与技术学院,西安710021

出　　处：《西安工业大学学报》2025年第1期149-161,共13页Journal of Xi’an Technological University

基　　金：陕西省教育厅自然专项项目(19JK0411)。

摘　　要：针对无人机等低小慢目标在复杂环境中感知与识别的准确性问题,采用一种基于多元声阵列的创新AI算法框架。该框架通过构建多任务学习模型,结合EfficientNet和GRU网络,利用灰狼优化算法(GWO)调整权重,显著提升了无人机型号识别和负载分析的准确性;提出一种基于MUSIC算法和DBSCAN聚类的信源定位方法,能够准确判定无人机的数量和位置。运用该方法设计了一种3D-GAA(GRU Acceleration Attention)轨迹预测网络,融合位置、速度和加速度信息,有效提高了轨迹预测的准确性和鲁棒性。实验结果表明,与传统方法相比,提出的方法在无人机识别准确率上提高了30%,达到75%,轨迹预测的平均绝对百分比误差(MAPE)从248.05降至10.89,尤其在处理复杂动态和数据模式时表现出显著优势。the accuracy of perception and identification of low,slowand small targets drones in complex environmentshis paper proposes an innovative AI algorithm framework based on a multivariate acoustic array.onstruct a multi task learning model that combines EfficientNet and GRU networksusing the Grey Wolf Optimization algorithm(GWO)to adjust weightssignificantly enhances the accuracy of drone model identification and payload analysis.Additionally,this paper introduces a source localization methodwhich can accurately determine the number and location of drones.3D-GAA(GRU Acceleration Attention)trajectory prediction network that integrates position,velocity,and acceleration information,effectively improving the accuracy and robustness of trajectory prediction.Experimental results show that compared to traditional methods,the proposed approach increases drone identification accuracy by 30%,reaching 75%,and reduces the Mean Absolute Percentage Error(MAPE)for trajectory prediction from 248.05 to 10.89,demonstrating significant advantages especially when handling complex dynamics and data patterns.

关 键 词：多元声阵 低小慢目标 AI算法 轨迹预测 

分 类 号：V279[航空宇航科学与技术—飞行器设计] E91[军事] TP18[自动化与计算机技术—控制理论与控制工程]