基于SAMME+ResNet的多相码信号识别方法  

Polyphase code signal recognition method based on SAMME+ResNet

作　　者：孙艺聪 田润澜[1] 董会旭 孙亮[2] SUN Yicong;TIAN Runlan;DONG Huixu;SUN Liang(School of Aviation Operations and Services, Aviation University of Air Force, Changchun 130022,China;Unit 93110 of the PLA, Beijing 100843, China)

机构地区：[1]空军航空大学航空作战勤务学院,吉林长春130022 [2]中国人民解放军93110部队,北京100843

出　　处：《系统工程与电子技术》2020年第10期2239-2245,共7页Systems Engineering and Electronics

基　　金：国家自然科学基金(61571462)资助课题。

摘　　要：针对传统多相码信号识别方法在低信噪比情况下分类精度不高、类识别率不均衡和识别方法不具有通用性的特点,提出了一种利用集成学习中的多类指数损失函数逐步添加模型(stagewise additive modeling using a multi-class exponential loss function,SAMME)算法和残差神经网络(residual neural network,ResNet)的多相码信号识别方法。通过仿真实验对5类多相码信号进行了分类识别,验证了模型的有效性,分析了不同数量基学习器对模型的影响,最后与传统分类方法进行了对比。仿真结果表明,在信噪比低于6 dB的情况下,所提方法相对于单个残差网络提高了约10%的分类精度,同时缩小了类之间识别率的差距,相对于常用的分类方法也有很大的优势。In view of the characteristics of the traditional polyphase code signal recognition methods,such as low classification accuracy,uneven class recognition rate and non-generality of recognition methods in the case of low signal to noise ratio(SNR),a polyphase code signal recognition method based on stagewise additive modeling using a multi-class exponential loss function(SAMME)algorithm in ensemble learning and residual neural network(ResNet)is proposed.Simulation experiments are carried out to classify and identify five kinds of polyphase code signals,and the validity of the model is verified.The influence of different quantity base learners on the model is analyzed.Finally,the proposed method is compared with the traditional classification methods.Simulation results show that when the SNR is lower than 6 dB,the proposed method improves the classification accuracy by about 10%compared with single residual network and reduces the difference of recognition rate between classes and also has great advantages over common classification methods.

关 键 词：多相码 信号识别 集成学习 残差神经网络 

分 类 号：TN971.1[电子电信—信号与信息处理]