摘要
针对区域重力似大地水准面与GPS/水准数据可能受系统误差影响而降低计算精度的问题,采用BP神经网络拟合区域似大地水准面,削弱系统误差等因素影响,提高区域似大地水准面的拟合效果。分析BP神经网络进行区域似大地水准面拟合的局限性。使用不同区域范围的数据进行验证,结果表明BP神经网络在控制区域内能够较好地逼近实际似大地水准面,而在区域之外具有较差的外推能力。
BP neural network is used to weaken the systematic errors in order to improve the accuracy of quasigeoid for the systematic errors in the data of GPS and level and the data of gravity quasigeoid may reduce the reliability.The deficiency for fitting the quasigeoid with BP neural network is discussed.Two sets of data in different areas are employed to test BP neural network.It is shown that BP neural network can factually approach the true quasigeoid,but have poor generalization ability out of control area.
出处
《测绘工程》
CSCD
2012年第1期21-24,共4页
Engineering of Surveying and Mapping
基金
863计划资助项目(2009AA121402-5)
陕测科(2009-03)
