基于遗传Elman神经网络进行矿区GPS高程拟合

目前,城市、平原地区的似大地水准面建立精度已经达到厘米级,但在矿区进行高程拟合时,由于地面高低起伏没有规则,其似大地水准面的拟合精度并不理想。针对此问题,本文提出利用遗传算法优化Elman神经网络的方法精化似大地水准面,采用移去-恢复法对残差进行建模,使用EGM 2008地球重力场模型和地形起伏信息来精化求解似大地水准面和参考椭球面之间的高程异常,同时着重分析了地球重力场模型以及地形变化信息对高程异常求解的重要性,并使用某矿区实测数据(GPS、水准)对所提方法进行验证,实验结果表明:文中所提方法的精度要优于二次曲面拟合模型和单一Elman模型,其外符合精度达到了1.14 cm,可以代替四等水准测量。

Using GPS Level to Refine the Quasi Geoid of Mining Area GPS Elevation Fitting of Mining Area Based on Genetic Elman Neural Network

At present,the accuracy of quasi-geoid establishment in cities and plains has reached centimeter level.However,the fitting precision of quasi-geoid is not ideal because of the irregular fluctuation of ground height in mining area.In order to solve this problem,presents a method to refine the quasi-geoid by using genetic algorithm to optimize Elman neural network,and to model the residual error in moving-restoring method.Elaboration of elevation anomalies between quasi-geoid and reference ellipsoid using EGM 2008 gravity field model and topographic relief information.At the same time,the emphatically analyzes the importance of the earth gravity field model and the terrain change information to the solution of the elevation anomaly,and verifies the proposed method by using the measured data(GPS,leveling)in a mining area.The experimental results show that the accuracy of the proposed method is better than that of the two-degree surface fitting model and the single Elman model.and its external compliance accuracy reaches 1.14 cm,which can replace fourth-order leveling.