北斗卫星钟差预报的BP神经网络法  

BP neural network method for Beidou satellite clock bias prediction

作　　者：朱宇航 李发利 李敬宇 ZHU Yuhang;LI Fali;LI Jingyu(School of Resources and Civil Engineering,Liaoning Institute of Science and Technology,Benxi,Liaoning 117004,China)

机构地区：[1]辽宁科技学院资源与土木工程学院,辽宁本溪117004

出　　处：《测绘标准化》2024年第4期29-34,共6页Standardization of Surveying and Mapping

基　　金：2024年辽宁科技学院大学生创新创业训练计划项目(202411430105)。

摘　　要：在北斗卫星钟差预报中,本文将粒子群优化算法引入到BP神经网络中,以提高北斗卫星钟差预报的精度。首先,利用粒子群算法优化BP模型的建模参数;其次,通过实验构建最优个体适应度值变化曲线图,优化前后BP模型训练误差;最后,对比分析不同模型的多次预报结果。结果表明,BP神经网络在经粒子群算法优化后北斗卫星钟差预报结果具有很好的精度和稳定性,其预报的精度和稳定性均优于二次多项式(QP)和GM(1,1)两种常用模型。This paper introduces particle swarm optimization algorithm into BP neural network,to improve the accuracy of Beidou satellite clock bias prediction.Firstly,the particle swarm optimization algorithm is used to optimize the modeling parameters of the BP model.Secondly,the variation curve of the optimal individual fitness value and the training error of the BP model before and after optimization are experimentally analyzed.Finally,the multiple prediction results of different models are compared and analyzed.The results show that the Beidou satellite clock bias prediction can provide good and stable prediction accuracy that BP neural network after particle swarm optimization,and its prediction accuracy and stability are better than the two commonly used models QP and GM(1,1).

关 键 词：卫星钟差预报 粒子群优化算法 BP神经网络 

分 类 号：P228[天文地球—大地测量学与测量工程]