基于GWO-SA-LSTM模型的调制识别算法  

Modulation Recognition Algorithm Based on GWO-SA-LSTM Model

作　　者：张逸凡 雷斌[2] 苏晨 刘光辉 ZHANG Yifan;LEI Bin;SU Chen;LIU Guanghui(School of Weapon Science and Technology,Xi’an Technological University,Xi’an 710021,China;Electronics and Information Engineering,Xi’an Technological University,Xi’an 710021,China)

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

出　　处：《自动化与仪表》2024年第5期1-5,9,共6页Automation & Instrumentation

摘　　要：针对传统自动调制识别方法对现代信号识别准确率较低的问题,该文提出了GWO-SA-LSTM模型,该模型通过自注意力机制增强长短期记忆网络(LSTM)关键特征捕获能力,并利用灰狼算法(GWO)优化超参数。在真实环境的调制识别数据集上的实验表明,该文模型在10 dB信噪比下达到了97%的最高准确率,整体表现优于单纯LSTM模型。Addressing the issue of traditional automatic modulation recognition methods’low accuracy for modern signals,this paper introduces the GWO-SA-LSTM model.This model enhances the capability of long short-term memory(LSTM)networks to capture key features through a self-attention mechanism and utilizes the grey wolf optimizer(GWO)to optimize hyperparameters.Experiments conducted on a real-world modulation recognition dataset demonstrate that the proposed model achieves a highest accuracy of 97%at a 10 dB signal-to-noise ratio,outperforming the standalone LSTM model in overall performance.

关 键 词：无线电监测 调制识别 长短期记忆网络 灰狼算法 自注意力机制 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TN911.3[自动化与计算机技术—控制科学与工程]