信号处理2025，Vol.41Issue(5)：829-839,11.DOI:10.12466/xhcl.2025.05.005
气象雷达鸟群目标多维特征建模与特性分析
Multidimensional Feature Modeling and Characteristic Analysis of Bird Flock Targets in Weather Radar
摘要
Abstract
With the expansion of civil aviation transportation and the improvement of ecological environments,bird strikes occur frequently,posing a threat to aviation safety.Birds are typical"low,slow,and small"targets,character-ized by low-altitude flight,slow speed,and small size.The core challenge in recognizing and classifying such targets lies in extracting meaningful features from data through deep feature learning and combining the physical characteristics of the target with a multidimensional feature model for comprehensive analysis.Meteorological radar is suitable for large-scale monitoring of migratory bird flocks.It can be effectively networked and is capable of acquiring various target characteristic parameters.Taking bird flocks and meteorological targets in the context of aviation safety as examples,we analyzed measured meteorological radar data.By parametrically representing radar multidimensional features,the char-acteristics of bird flocks,precipitation,and cloud targets were examined,with a focus on features such as target reflec-tivity,spectral width,and differential phase characteristics,among others.Through histogram visualization analysis,principal component analysis,and the Fisher linear discriminant method,significant statistical differences were identi-fied in key parameters—such as reflectivity and correlation coefficient—demonstrating their effectiveness in distinguish-ing bird flocks from meteorological targets.The experimental results show that the correlation coefficient of bird flock targets presents a single-peak normal distribution,the spectrum width distribution is relatively dispersed,the reflectivity is highly concentrated in the 10～20 dBZ interval,and the differential phase is concentrated in the high value interval of 75°～125°,all of which differ significantly from meteorological targets.By clustering different target types and calculat-ing the two-norm distance between targets,further evidence of significant separability between bird flocks and meteoro-logical targets is provided.These findings offer both data-driven and theoretical support for developing a bird-strike warning system.关键词
气象雷达/回波特征/特性分析/鸟击预警Key words
weather radar/echo characteristics/characteristic analysis/bird strike warning分类
信息技术与安全科学引用本文复制引用
张梦晨,赵志坚,刘佳,陈小龙,苏添记,房旭东,薛伟,关键..气象雷达鸟群目标多维特征建模与特性分析[J].信号处理,2025,41(5):829-839,11.基金项目
国家自然科学基金(62222120) （62222120）
国家重点研发计划(2024YFB3909800) （2024YFB3909800）
山东省自然科学基金(ZR2024JQ003) The National Natural Science Foundation of China(62222120) （ZR2024JQ003）
National Key Research and Development Program of China(2024YFB3909800) （2024YFB3909800）
Shandong Provincial Natural Science Foundation(ZR2024JQ003) （ZR2024JQ003）
