BDS navigation satellite clock difference prediction based on PSO-Elman neural network
-
摘要: 卫星钟差是导航定位系统定位精度的重要影响因子之一. 针对北斗卫星导航系统(BDS)精密钟差预报性能寻优问题，提出一种基于粒子群优化 (PSO) 算法优化Elman神经网络的钟差预报模型方法(PSO-Elman模型)，以解决Elman神经网络局部最优问题对钟差预报结果的影响. 首先对钟差产品进行预处理；然后通过PSO算法迭代寻优确定Elman神经网络权值、阈值的初始值，并将进行预处理之后的序列数据进行训练建模；再采用武汉大学国际GNSS服务(IGS)数据分析中心(WHU)提供的BDS精密钟差产品数据进行钟差预测；最后将预测结果还原为预报钟差. 结果表明：对比于二次多项式(QP)模型、附加周期项多项式(SA)模型、灰色(GM)模型，PSO-Elman模型精度分别提高90. 7%、84. 2%、81. 6%，稳定度提高85. 3%、76. 3%、36. 1%. 实验表明：PSO-Elman模型在1~12 h短期预报模拟结果的预报精度和稳定性有显著提高，验证了提出方法的可行性.
-
关键词:
- 北斗卫星导航系统(BDS) /
- 钟差预报 /
- 粒子群优化(PSO)算法 /
- Elman神经网络 /
- 卫星钟差
Abstract: Satellite clock error is one of the important factors affecting the positioning accuracy of navigation and positioning system. Aiming at the problem of optimizing the precision clock error prediction performance of the BeiDou Navigation Satellite System (BDS), a method of optimizing the Elman neural network clock error prediction model based on particle swarm optimization (PSO) is proposed to solve the influence of the local optimal problem of Elman neural network on the clock error prediction results. Firstly, the clock error product is preprocessed. The initial weights and thresholds of Elman neural network are determined by iterative optimization of PSO algorithm, and the preprocessed sequence data are used for training modeling. The BDS precision clock error product data provided by IGS Data Analysis Center (WHU) of Wuhan University are used to predict the clock error, and then the prediction results are restored to predict the clock error. The results show that compared with the quadratic polynomial (QP) model, the polynomial (SA) model with additional period term, and the grey (GM) model, the accuracy is improved by 90.7%, 84.2%, 81.6%, and the stability is improved by 85.3%, 76.3%, 36.1%, respectively. The experimental results show that the prediction accuracy and stability of PSO-Elman model are significantly improved in 1−12 h short term forecast simulation, which verifies the feasibility of the proposed method. -
表 1 C46号卫星钟差预报结果统计
ns C46
钟差/hPSO-Elman QP SA GM RMSE Range RMSE Range RMSE Range RMSE Range 1 0.022 7 0.069 9 0.226 0.440 0.029 3 0.051 7.16 1.68 3 0.059 6 0.110 0 0.232 0.607 0.113 0 0.150 7.70 5.07 6 0.151 0 0.457 0 0.183 0.637 0.343 0 0.414 9.03 10.10 12 0.223 0 0.886 0 1.730 3.940 2.080 0 4.280 11.40 20.00 
下载: 导出CSV
表 2 3颗卫星四种模型钟差预报结果统计
ns 预测卫星 PSO-Elman QP SA GM RMSE Range RMSE Range RMSE Range RMSE Range C19 0.279 0.660 3.27 5.74 1.16 1.88 0.449 0.862 C38 0.391 0.975 6.15 9.72 3.13 6.40 0.833 0.879 C46 0.223 0.886 1.73 3.94 2.08 4.28 11.400 20.000
下载: 导出CSV
-
[1] WANG Y P, LU Z P, QU Y Y, et al. Improving prediction performance of GPS satellite clock bias based on wavelet neural network[J]. GPS solutions, 2017, 21(2): 523-534. DOI: 10.1007/s10291-016-0543-z [2] 熊红伟, 程新文, 张海涛, 等. 卫星钟差单差的小波神经网络预报[J]. 测绘科学, 2017, 42(9): 9-14. [3] YANG Y X, XU Y Y, LI J L, et al. Progress and performance evaluation of BeiDou global navigation satellite system: Data analysis based on BDS-3 demonstration system[J]. Science china earth sciences, 2018, 61(5): 614-624. DOI: 10.1007/s11430-017-9186-9 [4] HUANG G W, CUI B B, ZHANG Q, et al. An improved predicted model for BDS ultra-rapid satellite clock offsets[J]. Remote sensing, 2018, 10(2): 60. DOI: 10.3390/rs10010060 [5] 李超, 党亚民, 谷守周. 灰色模型的GLONASS卫星钟差预报[J]. 导航定位学报, 2016, 4(4): 24-29. DOI: 10.16547/j.cnki.10-1096.20160405 [6] 艾青松, 徐天河, 孙大伟, 等. 顾及周期项误差和起点偏差修正的北斗卫星钟差预报[J]. 测绘学报, 2016, 45(2): 132-138. [7] 王宇谱, 吕志平, 陈正生, 等. 卫星钟差预报的小波神经网络算法研究[J]. 测绘学报, 2013, 42(3): 323-330. [8] XI C, CAI C L, LI S M. et al. Long-term clock bias prediction based on an ARMA model[J]. Chinese astronomy and astrophysics, 2014, 38(3): 342-354. DOI: 10.1016/j.chinastron.2014.07.010 [9] 押少帅, 赵兴旺, 胡豪杰, 等. 北斗三号卫星钟差短期预报与稳定性分析[J]. 全球定位系统, 2021, 46(3): 39-46. DOI: 10.12265/j.gnss.2020120703 [10] 王润, 王井利, 吕栋. 导航卫星钟差预报的Elman神经网络算法研究[J]. 大地测量与地球动力学, 2021, 41(3): 285-289. DOI: 10.14075/j.jgg.2021.03.012 [11] 程博, 裘晓枫, 季凌燕, 等. BDS精密卫星钟差建模与预报[J]. 测绘科学, 2019, 44(10): 1-9. DOI: 10.16251/j.cnki.1009-2307.2019.10.003 [12] 李特. BDS星载原子钟性能分析及精密钟差建模预报[D]. 辽宁: 辽宁工程技术大学, 2021. [13] 马昌忠, 王潜心, 胡超, 等. 基于BDS-2/BDS-3联合处理的北斗超快速钟差预报优化策略[J]. 导航定位与授时, 2020, 7(5): 28-36. [14] 王宇谱, 吕志平, 王宁. BDS星载原子钟长期性能分析[J]. 测绘学报, 2017, 46(2): 157-169. DOI: 10.11947/j.AGCS.2017.20160369 [15] 戴文战, 娄海川, 杨爱萍. 非线性系统神经网络预测控制研究进展[J]. 控制理论与应用, 2009, 26(5): 521-530. [16] 吕栋, 欧吉坤, 于胜文. 基于MEA-BP神经网络的卫星钟差预报[J]. 测绘学报, 2020, 49(8): 993-1003. DOI: 10.11947/j.AGCS.2020.20200002 [17] XU B, WANG Y, YANG X H. Navigation satellite clock error prediction based on functional network[J]. Neural processing letters, 2013, 38(2): 305-320. DOI: 10.1007/s11063-012-9247-8 [18] WANG S L, LIU G Y, GAO M, et al. Heterogeneous comprehensive learning and dynamic multi-swarm particle swarm optimizer with two mutation operators[J]. Information sciences, 2020, 540(3): 175-201. DOI: 10.1016/j.ins.2020.06.027 [19] WANG J, JU C W, GAO Y, et, al. A PSO based energy efficient coverage control algorithm for wireless sensor networks[J]. Computers, materials and continua, 2018, 56(3): 433-446. DOI: 10.3970/cmc.2018.04132 [20] 王宇谱, 吕志平, 崔阳, 等. 利用遗传小波神经网络预报导航卫星钟差[J]. 武汉大学学报(信息科学版), 2014, 39(7): 809-914. [21] 张雨浓, 肖秀春, 陈扬文. Hermite前向神经网络隐节点数目自动确定[J]. 浙江大学学报:工学版, 2010, 44(2): 271-275. -
点击查看大图
图(5) / 表(2)
计量
- 文章访问数: 277
- HTML全文浏览量: 89
- PDF下载量: 25
- 被引次数: 0
