变化地磁场预测的支持向量机建模  

Modeling and forecasting of the variable geomagnetic field by support vector machine

作　　者：易世华 刘代志[1] 何元磊 齐玮 

机构地区：[1]第二炮兵工程大学,西安710025

出　　处：《地球物理学报》2013年第1期127-135,共9页Chinese Journal of Geophysics

基　　金：国家自然科学基金(40974037)资助

摘　　要：变化地磁场建模与预测是地磁导航、空间环境监测等领域的重要研究课题.由于变化地磁场属于日地系统中的一部分,受多种因素的制约影响,且其变化本身也具有较强的前后相关性.本文综合空间和地面监测数据,以变化地磁场地面观测数据、地方时、太阳射电流量和行星际磁场南向分量等为输入,采用支持向量机方法,建立了变化地磁场综合模型,并进行预测.结果表明,在地磁活动Kp指数小于4时,预测3h平均绝对误差小于1.61nT.Modeling and forecasting of the variable geomagnetic field is an important research topic in geomagnetic navigation and space environment monitoring. The variable geomagnetic field belongs to the sun-earth system, it depends on a variety of factors, and the present value of the variable geomagnetic field is related to the past. In this paper, we established the comprehensive model of variable geomagnetic field by the support vector machines. The model combines the space and ground monitoring data, and it uses the ground measuring data of the variable geomagnetic field, local time, the solar radio flux, the southern component of the interplanetary magnetic field as input. Experiments show that the mean absolute error of the model forecasting three hours are less than 1.61 nT when K9 is less than 4.

关 键 词：变化地磁场 地磁导航 空间环境监测 支持向量机 建模 预测 

分 类 号：P318[天文地球—固体地球物理学]