多面函数参数自适应选取方法在GPS高程拟合中的应用  

Application of Self-Adaptive Parameter Selection for Multiquadric Function in GPS Elevation Fitting

作　　者：谭江涛 王章朋 钟波[1] 丁剑[3] TAN Jiangtao;WANG Zhangpeng;ZHONG Bo;DING Jian(School of Geodesy and Geomatics,Wuhan University,129 Luoyu Road,Wuhan 430079,China;Shanghai Baoye Group Co Ltd,77 Qing’an Road,Shanghai 200941,China;Chinese Academy of Surveying and Mapping,28 West-Lianhuachi Road,Beijing 100830,China)

机构地区：[1]武汉大学测绘学院,武汉市430079 [2]上海宝冶集团有限公司,上海市200941 [3]中国测绘科学研究院,北京市100830

出　　处：《大地测量与地球动力学》2020年第8期832-837,共6页Journal of Geodesy and Geodynamics

基　　金：国家重点研发计划(2018YFF0300202);国家自然科学基金(41974015);民用航天技术预研项目(D010103)。

摘　　要：提出一种综合的多面函数参数自适应选取方法。该方法利用核函数阶数和平滑因子存在最优值以及核函数结点分布应符合高程异常变化趋势的特点,综合利用正交化算法选取核函数结点以及二维粒子群算法确定最佳阶数和平滑因子,实现多面函数参数的完全自适应选取。将该方法应用于地形起伏差异不同的2个测区,结果表明,相比传统经验性方法和部分参数自适应方法,该方法用于GPS高程拟合的精度和可靠性更高。We propose an integrated self-adaptive parameter selection scheme.Our method takes advantage of the fact that the kernel function order and the smoothing factor are in an optimal range,and the distribution of kernel function nodes should be in accordance with terrain characteristics.By combining the orthogonalization algorithm to select the nodes and the two-dimensional particle swarm algorithm to determine the optimal order and smoothing factor,we realize the fully self-adaptive selection of multiquadric function parameters.Two areas with different topographic relief are selected for GPS elevation fitting;the results show that accuracy and reliability of our method are both better than the traditional empirical parameter selection and partially self-adaptive parameter selection method.

关 键 词：多面函数法 GPS高程拟合 自适应参数选取 正交化算法 二维粒子群算法 

分 类 号：P228[天文地球—大地测量学与测量工程]