基于LM-BP神经网络的GPS高程拟合算法研究  

Study on GPS Elevation Fitting Algorithm Based on LM-BP Neural Network

作　　者：和会[1] 闻洪峰[1] 魏太渊[1] 

机构地区：[1]国家测绘局重庆测绘院,重庆400015

出　　处：《测绘与空间地理信息》2011年第6期95-97,101,104,共5页Geomatics & Spatial Information Technology

摘　　要：提出了基于LM-BP神经网络方法转换的GPS高程拟合算法,并与传统拟合方法进行了比较。经实例验证,基于LM-BP神经网络的GPS高程拟合优于传统拟合方法,特别是在拟合点较少的大范围GPS拟合中显著改善了拟合高程的精度,具有一定的实用价值。The algorithm of GPS elevation fitting based on LM-BP neural network is presented and compared with traditional fitting methods.The GPS leveling experiment testified that this GPS elevation fitting algorithm is better than traditional methods,especially improves the elevation precision distinctly in GPS fitting with few fitting points in a large area.It has certain practical value.

关 键 词：神经网络 高程拟合 LM-BP算法 过度拟合 

分 类 号：P228.4[天文地球—大地测量学与测量工程]