一种基于智能算法的GNSS高程拟合方法  

A GNSS Elevation Fitting Method Based on Intelligent Algorithm

作　　者：王朝[1] 王志文 Wang Chao;Wang Zhiwen(CCCC First Harbor Consultants Co.,Ltd.,Tianjin 300222,China)

机构地区：[1]中交第一航务工程勘察设计院有限公司,天津300222

出　　处：《港口航道与近海工程》2024年第3期86-90,共5页Port,Waterway and Offshore Engineering

摘　　要：广义回归神经网络(GRNN)是一种新型的前馈神经网络模型,具有训练次数少、耗时短、非线性参数的预报能力较强等优点。但GRNN唯一的调节参数SPREAD不能自动获取限制其进一步的应用。针对该缺陷,本文采用果蝇优化算法(FOA)与GRNN相结合构建FOAGRNN模型对GRNN进行优化,自动获取调节参数的值。为了检验FOAGRNN模型的GNSS高程拟合精度,进行了实验分析。实验结果证明了FOAGRNN模型的GNSS高程拟合精度可达6mm。为进一步检验FOAGRNN模型的优越性,采用与平面拟合模型、二次曲面拟合模型进行对比。实验结果表示FOAGRNN模型的拟合精度要优于平面拟合模型和二次曲面拟合模型,证明了FOAGRNN模型在数据样本较少的情况下,其GNSS高程拟合精度仍然可以达到较高精度。Generalized regression neural network(GRNN)is a new feed-forward neural network which has many advantages such as fewer training times,short period,strong forecasting capability of nonlinear parameters and etc.However,as the only adjustable parameter of GRNN,SPREAD cannot be obtained automatically,which limits its further application.To solve this drawback,Fly Optimization Algorithm(FOA)is combined with GRNN to build FOAGRNN model,which optimizes GRNN model and achieves automatic gathering of adjustable parameter.In order to test the accuracy of GNSS elevation fitting based on FOAGRNN model,an experimental analysis is carried out.The results show that the above accuracy of GNSS height fitting reaches 6 mm.FOAGRNN model is also compared with plane fitting model and quadric fitting model,which shows that the superiority of FOAGRNN model in fitting accuracy.In conclusion,FOAGRNN model supports higher accuracy of GNSS height fitting even though less data samples are available.

关 键 词：果蝇优化算法(FOA) 广义回归神经网络(GRNN) GNSS高程拟合 

分 类 号：P216[天文地球—测绘科学与技术]