摘要
为了提升卫星钟差预报的精确性,推动其在卫星导航定位系统中的广泛应用,本文提出一种新的循环钟差预报模型。该模型结合了改进粒子群优化(IPSO)算法和反向传播(BP)神经网络模型,实现对神经网络模型参数的高效优化。首先,借助IPSO算法优化的BP神经网络模型,对卫星的初始钟差序列进行高精度的预报;其次,采用自回归(AR)模型,精确校正IPSO-BP模型预报中产生的误差;最终,将AR模型的校正结果与IPSO-BP模型的预报结果相结合,形成更为精准的预报结果。为了验证该模型的性能,选取了4种典型的卫星钟差序列作为测试样本。实验结果显示,本文提出的组合预报模型在6 h的卫星钟差预报中,稳定性和精度上较BP神经网络模型、AR模型及IPSO-BP模型分别提高了78.63%、33.04%、29.25%和84.13%、52.38%、40.00%,实验成果验证了本文所提出模型的优越性能。
To enhance the accuracy of satellite clock offset prediction and promote its widespread application in satellite navigation and positioning systems,this paper proposes a novel cyclic clock offset prediction model.This model integrates the improved particle swarm optimization(IPSO)algorithm and the back propagation(BP)neural network model to achieve efficient optimization of the neural network model's parameters.Firstly,the BP neural network model optimized by the IPSO algorithm is employed to predict the initial satellite clock offset sequence with high accuracy.Secondly,an autoregressive(AR)model is adopted to accurately correct the errors generated by the IPSOBP model prediction.Finally,the corrected results of the AR model are combined with the prediction results of the IPSO-BP model to form a more accurate prediction.To validate the performance of this model,four typical satellite clock offset sequences are selected as test samples.The experimental results show that the proposed combined prediction model improves the stability and accuracy by 78.63% and 84.13% compared to the BP neural network model,by 33.04% and 52.38% compared to the AR model,and by 29.25% and 40.00% compared to the IPSO-BP model,respectively,in 6-hour satellite clock offset prediction.These results verify the superior performance of the proposed model.
作者
赵妮妮
ZHAO Nini(Heilongjiang Institute of Geomatics Engineering,Harbin 150081,China)
出处
《测绘与空间地理信息》
2025年第S1期82-85,共4页
Geomatics & Spatial Information Technology
关键词
卫星钟差预报
BP神经网络
改进粒子群优化算法
自回归模型
satellite clock offset prediction
BP neural network
improved particle swarm optimization algorithm
autoregressive model
作者简介
赵妮妮(1992-),女,辽宁海城人,助理工程师,学士,主要从事遥感测绘等方面的应用研究工作。
