摘要
为使光缆维护管理部门对未来光缆线路传输损耗进行科学预测,以灰色GM(1,1)模型、三次指数平滑模型和支持向量回归(SVR)模型为基础,构建权重随时间变化的组合预测模型,并进行了实例研究及Matlab仿真。仿真结果表明,SVR模型在三个单项模型中预测效果最优,且变权组合模型无论在拟合期还是检验期,其预测效果都优于其它单项模型。
In order to make the department of optical cable maintenance and management make scientific predictions on the transmission loss of the future optical cable,based on the grey GM(1,1) model,the three-exponential smoothing model and the support vector regression(SVR) model,the paper builds a combined forecasting model for the weights change over time,and conducted an example study and Matlab simulation.The simulation results show that the SVR model has the best forecast effect among the three individual models,and the variable weight combination model is superior to other single-term models in both the fitting period and the testing period.
作者
张明江
李红卫
夏贵进
ZHANG Mingjiang, LI Hongwei, XIA Guijin(Experimental training base of institute of information and communication of national university of defense technology, Xi'an 710106, Chin)
出处
《光通信技术》
北大核心
2018年第5期45-49,共5页
Optical Communication Technology
关键词
光缆传输损耗预测
变权组合模型
支持向量回归
GM(1
1)
三次指数平滑
transmission loss prediction of optical cable
variable weight combination model
support vector regression
GM(1,1)
three exponential smoothing