本文论述了最小二乘支持向量机(LS-SVM)的算法,提出了基于最小二乘支持向量机进行GPS高程拟合的方法,并在MATLAB中编制了相应的LS-SVM程序,建立了相应的GPS高程拟合模型。以实例数据讨论了LS-SVM的GPS高程拟合的分析方法,通过与多项式拟合、BP神经网络拟合、GA-BP神经网络拟合的结果比较,可知LS-SVM的拟合精度较高。
This paper discussed the algorithm of support vector regression, proposed the method of GPS leveling based on the least squares support vector machine, and compiled programs for GPS leveling by using the MATLAB toolbox. Through a set of experimental data, the effectiveness of our method was discussed. Compared with results of polynomial fitting, BP neural network fitting and GA-BP neural network fitting, the result of LS-SVM is more promising.