自适应加权混合集成学习的雷达工作模式识别方法
Method for Radar Working Mode Recognition Based on Adaptive Weighted Hybrid Ensemble Learning
王之腾 1纪存孝 1刘畅 1王润雪 1王水斌2
作者信息
- 1. 陆军工程大学 通信工程学院,江苏 南京 210007
- 2. 杭州电子科技大学 通信工程学院,浙江 杭州 310000
- 折叠
摘要
为提高在现代战场中的生存能力,机载多功能雷达呈现出信号样式复杂、参数高度捷变、开关机无规律和辐射时间减少等特点,给基于传统方法的雷达工作模式识别带来了一定的挑战.参考多功能雷达常见工作模式典型特征参数生成的样本数据,基于水波中心扩散(water wave center diffusion,WWCD)算法优化多个模型参数,采用自适应加权策略提高多个模型集成学习算法性能,开展多功能雷达工作模式识别.实验分别使用遗传算法、粒子群优化算法、差分进化算法和WWCD算法优化单模型参数,使用软投票、硬投票、自适应加权等不同集成学习策略进行工作模式识别.结果表明,所提算法相较于传统算法具有较高的准确率.此外,还研究测试了该算法在小样本条件下识别雷达工作模式的性能,验证了该算法的可行性和较高的识别效率.
Abstract
To enhance survival capabilities in modern battlefields,airborne multifunction radars exhibit the characteristics such as complex signal patterns,highly variable parameters,irregular on/off switching,and reduced radiation times,posing certain challenges to radar operating mode recognition based on tradi-tional methods.By referring to sample data generated from typical characteristic parameters of common operating modes of multifunction radars,multiple model parameters are optimized with the water wave center diffusion(WWCD)algorithm,and an adaptive weighting strategy is employed to improve the per-formance of ensemble learning algorithms for multifunction radar operating mode recognition.Experi-ments were conducted using genetic algorithms,particle swarm optimization algorithms,differential evo-lution algorithms,and the WWCD algorithm to optimize single model parameters,and different ensemble learning strategies such as soft voting,hard voting,and adaptive weighting were used for operating mode recognition.The results demonstrate that the proposed algorithm achieves higher accuracy,compared with traditional algorithms.Furthermore,the performance of the algorithm in recognizing radar operating modes under small sample conditions was also tested,verifying the feasibility and high recognition efficiency of this algorithm.
关键词
水波中心扩散算法/自适应加权集成学习/小样本/雷达工作模式识别Key words
water wave center diffusion(WWCD)algorithm/adaptive weighted ensemble learning/small sample/radar working mode recognition引用本文复制引用
出版年
2024
NETL
NSTL
万方数据