自适应卡尔曼滤波的电离层TEC预测模型改进  

Improvement of prediction model for ionospheric TEC with adaptive Kalman filter

作　　者：王建敏[1] 黄佳鹏[1] 刘梓然 祝会忠[1] 马天明[1] WANG Jianmin;HUANG Jiapeng;LIU Ziran;ZHU Huizhong;MA Tianming(School of Geomatics, Liaoning Technical University, Fuxin, Liaoning 123000, China;School of Civil Engineering, Liaoning University of Science and Technology, Anshan, Liaoning 114000, China)

机构地区：[1]辽宁工程技术大学测绘与地理科学学院,辽宁阜新123000 [2]辽宁科技大学土木工程学院,辽宁鞍山114000

出　　处：《导航定位学报》2018年第2期121-127,共7页Journal of Navigation and Positioning

基　　金：国家自然科学基金项目(41474020;41504010);国家863计划项目(2014AA121501);辽宁省博士启动基金支持项目(20141141)

摘　　要：针对直接使用IGS公布的电离层总电子数进行建模会导致预测模型建立存在偏差,以及使用传统卡尔曼滤波在对大量数据进行预处理时容易导致数据发散,进而降低电离层TEC模型预测精度的问题,提出一种利用自适应卡尔曼滤波的改进方法,使用方差补偿自适应卡尔曼滤波对原始数据进行预处理,再利用小波神经网络完成预测,最后分析模型预报的精准度。实验结果表明,此方法的预测平均精度相对直接使用原始数据建模和传统卡尔曼滤波都有不同程度的提高。Aiming at the problems that it is vulnerable to bias for the prediction model with the total number of electrons（TEC）directly obtained from IGS,and it is liable to data divergence for traditional Kalman filtering in the preprocess of large amounts of data,which lead to reduce the accuracy of predition model for ionospheric TEC,the paper proposed an improvement method using self-adaptive Kalman filtering：the self-adaptive Kalman filtering with variance compensation was used to preprocess the original data,and the wavelet neural network was used to accomplish the prediction,finally the prediction accuracy of the model was given.Experimental result showed that the average prediciton accuracy of the method would be higher both than that of modeling directly using original data and that of traditional Kalman filtering.

关 键 词：自适应卡尔曼滤波 方差补偿 小波神经网络 电离层总电子数 

分 类 号：P228[天文地球—大地测量学与测量工程]