一种优化的基于神经网络的经验ZTD模型

发布时间：2018-07-07 10:30

本文选题：ZTD + 神经网络模型　；参考：《测绘通报》2017年01期

【摘要】：目前,经验对流层天顶延迟(ZTD)模型已经有了飞速的发展,因为它们在使用时无需任何测量的实时地面气象数据,这给GNSS用户提供了极大方便。神经网络技术在实测参数型的ZTD建模中已经取得了一定的成果。与此同时,国内虽然有学者构建了神经外网络的经验ZTD模型,其最大的缺点是忽略了ZTD时间变化且只能单独预报ZTD。本文针对这些缺点构建了优化的神经网络经验ZTD模型。试验结果表明,本文提出的神经网络模型可以分别预报天顶干延迟ZHD和天顶湿延迟ZWD,且具有良好的精度:ZHD的Bias和RMSE分别为-3.7和19.8 mm;ZWD的Bias和RMSE分别为-0.6和34.2 mm。本文的神经网络模型预报的ZHD和ZWD的精度均与目前世界著名的GPT2w格网模型相当。另外,与GPT2w模型相比较,神经网络模型最大的优点就是无需庞大的预存格网数据作为输入,在使用时仅需要知道一个训练好的神经网络即可,该特点为GNSS用户提供了极大的方便。
[Abstract]:At present, empirical tropospheric zenith delay (ZTD) models have been rapidly developed because they do not require any real time ground meteorological data, which provides great convenience to GNSS users. Neural network technology has achieved some results in the ZTD modeling of measured parameters. At the same time, although some domestic scholars have constructed the empirical ZTD model of the neural network, its biggest disadvantage is to ignore the time variation of the ZTD and to predict the ZTD alone. In this paper, an optimized empirical ZTD model of neural network is constructed for these shortcomings. The experimental results show that the proposed neural network model can predict ZHD and ZWDs, and the Bias and RMSE of ZWD with good accuracy are -3.7 and 19.8mm respectively, and the bias and RMSE of ZWD are -0.6 and 34.2 mm. respectively. The precision of ZHD and ZWD predicted by the neural network model in this paper is similar to that of the world famous GPT2w grid model. In addition, compared with the GPT2w model, the biggest advantage of the neural network model is that it does not need large pre-stored grid data as input, and it only needs to know a trained neural network to use it. This feature provides great convenience for GNSS users.
【作者单位】：东南大学交通学院;
【基金】：国家自然科学基金(41574022;41274028) 普通高校研究生科研创新计划(JSCHKY201314)
【分类号】：P228.4

【相似文献】