L-PLC信道特性预测的SVM实现  

SVM Algorithm for L-PLC Channel Characteristics Prediction

作　　者：王振朝[1] 甘玉涛[1] 侯惠然 张士兵[1] 

机构地区：[1]河北大学电子信息工程学院,保定071002 [2]中电飞华通信有限公司,北京100029

出　　处：《电力系统及其自动化学报》2009年第4期36-40,共5页Proceedings of the CSU-EPSA

基　　金：河北省科技攻关资助项目(06213552);河北大学省部共建基金资助项目

摘　　要：信道特性预测是低压电力线载波通信L-PLC(low-vlotage power line communication)信道特性研究的重要组成部分。文中基于低压电力线信道的混沌特性,将电力线上的混沌时间序列映射到高维特征空间实现非线性映射,基于支持向量机SVM(support vector machine)算法建立了L-PLC信道特性预测模型,并给出了具体实现步骤。以最小均方误差MSE(mean square error)为标准,总结了SVM算法参数与模型预测性能以及参数间的关系。通过与径向基函数RBF(radial basis function)网络预测结果相比较,证明了基于SVM的L-PLC信道特性预测模型的有效性。Channel characteristics prediction is an important part of the research on low-vlotage power line communication （L-PLC）channel characteristics. According to the conclusion that L-PLC is chaotic, the chaotic time series measured from L PLC are mapped into high dimensional feature space. Based on support vector machine （SVM） algorithm, this paper estabilishes the prediction model of L-PLC channel characteristics and gives the concrete steps of prediction. With the least mean square error （MSE） principle, the relations between the parameters of SVM algorithm and the prediction performance and the relations Between the parameters are summarized. Compared with radial basis function （RBF） networks, the prediction model based on SVM is proved to be effectine.

关 键 词：信道特性 预测 支持向量机 径向基函数 

分 类 号：TM73[电气工程—电力系统及自动化] TP391.41[自动化与计算机技术—计算机应用技术]