作者：董重重¹, 何行¹, 孙秉宇¹, 谢玮¹, 蔡兵兵², 王先培²

作者单位：1. 国网湖北省电力有限公司计量中心, 湖北 武汉 430000;
2. 武汉大学电子信息学院, 湖北 武汉 430000

关键词：低压电力线载波通信;调制信号识别;特征参数集;支持向量机

摘要：

针对当前对低压电力线载波通信调制信号识别过程中出现选取特征困难、选取特征不恰当、识别准确率低的问题，提出特征选择——支持向量机（support vector machine， SVM）的调制信号识别方法。通过采集电力线载波通信芯片发送的调制信号样值，经预处理去噪、滤波后选取调制信号多个特征，使用特征选择工具——FEAST，从多个特征集中找出最能标识数据特征集的特征子集，利用SVM方法对特征子集进行判决归类，并将分类识别后的结果与传统神经网络进行比较。仿真结果表明，所提出的方法选取特征与原有方法相比更为简单准确，其识别准确率较传统神经网络有明显提升，调制信号识别准确率达到98%以上，且收敛速度相比更快，可为多特征下低压电力线载波通信调制信号识别提供参考。

Research on carrier signal modulation signal recognition method based on support vector machine
DONG Chongchong¹, HE Xing¹, SUN Bingyu¹, XIE Wei¹, CAI Bingbing², WANG Xianpei²
1. Metering Center of State Grid Hubei Electric Power Co., Ltd., Wuhan 430000, China;
2. College of Electronic Information, Wuhan University, Wuhan 430000, China
Abstract: Aiming at the problem that the selection characteristics of the low-voltage power line carrier communication modulation signal are difficult, the selection features are inappropriate, and the recognition accuracy is low, the feature selection-SVM support vector machine modulation signal identification method is proposed. By collecting the modulated signal samples sent by the power line carrier communication chip, after pre-processing denoising and filtering, selecting multiple features of the modulated signal, using the feature selection tool FEAST, the most characteristic data feature set can be found from multiple feature sets. The feature subsets are classified by the SVM support vector machine method, and the results of the classification and recognition are compared with the traditional neural network. The simulation results show that the proposed method is simpler and more accurate than the original method, and its recognition accuracy is significantly improved compared with the traditional neural network. The accuracy of modulation signal recognition is over 98%, and the convergence speed is faster. It can provide reference for multi-feature low-voltage power line carrier communication modulation signal identification.
Keywords: low-voltage power line carrier communication;modulation signal identification;feature parameter set;SVM
2019, 45(11):101-107  收稿日期: 2019-01-13;收到修改稿日期: 2019-04-08
基金项目: 国家自然科学基金资助项目（51707135）；国网湖北省电力有限公司电力科学研究院外委研究项目（HB1842）
作者简介: 董重重(1985-),男,湖北钟祥市人,工程师,硕士,研究方向为计量采集
参考文献
[1] 刘科满, 邓炜昱, 李利品, 等. 基于BPSK的低压电力线载波通信系统研究[J]. 现代电子技术, 2018, 41(7):16-20
[2] 王毅, 段大程, 程昊, 等. 基于神经网络的低压电力线载波通信信号调制识别研究[J]. 电测与仪表, 2016, 53(11):51-55
[3] 刘丹, 马秀荣, 单云龙. 一种基于ST-RFT算法的数字调制信号识别方法[J]. 计算机科学, 2018, 45(5):64-75
[4] 董国华, 朱莉乐, 李璟健, 等. 粗糙集结合神经网络的哮喘辨证分型法[J]. 计算机与数字工程, 2015, 43(9):1622-1626
[5] 王兰勋, 孟祥雅, 佟婧丽. 基于信噪比估计和盒维数的调制信号识别[J]. 河北大学学报(自然科学版), 2014, 34(6):656-661
[6] 袁兴梅, 杨明, 杨杨. 一种面向不平衡数据的结构化SVM集成分类器[J]. 模式识别与人工智能, 2013, 26(3):315-320
[7] 周长林, 钱志升, 王勤民, 等. 基于PSO-SVM方法的电源线传导泄漏信号识别与还原[J]. 电子与信息学报, 2018, 40(9):2206-2211
[8] 王洪智, 刘震, 李东辉. 基于多分类支持向量机的网络流量预测方法[J]. 科技导报, 2014, 32(17):60-63
[9] 李素姝, 王士同, 李滔. 基于LS-SVM与模糊补准则的特征选择方法[J]. 山东大学学报(工学版), 2017, 47(3):34-42
[10] 江伟华, 曹秀岭, 童峰. 采用支持向量机的水声通信信号调制识别方法[J]. 厦门大学学报(自然科学版), 2015, 54(4):534-539
[11] 江伟华. 水声通信信号调制识别及其实现研究[D].厦门:厦门大学, 2016.
[12] 张初. 基于光谱与光谱成像技术的油菜病害检测机理与方法研究[D]. 杭州:浙江大学, 2016.
[13] TARAN S, BAJAJ V, SHARMA D. TEO separated AM-FM components for identification of apnea EEG signals[C]//IEEE International Conference on Signal & Image Processing. IEEE, 2017.
[14] HASSANPOUR S, PEZESHK A M, BEHNIA F. Automatic digital modulation recognition based on novel features and support vector machine[C]//International Conference on Signal-image Technology & Internet-based Systems. IEEE, 2017.