基于支持向量机的GPS似大地水准面拟合

GPS高精度平面控制成果在各类工程中已经得到了广泛应用,然而其高程信息目前正在作进一步的研究.介绍统计学习理论和支持向量机,提出利用支持向量机技术进行似大地水准面拟合.以实测GPS定位数据为试验资料,对支持向量机和神经网络以及多项式拟合的结果进行比较分析:支持向量机技术拟合数据的精度达到了神经网络和多项式拟合的精度,并且解决了神经网络技术不能实时处理数据、过学习、收敛速度慢、易陷于局部极值等问题.

GPS Quasi Geoid Fitting Based on Support Vector Machine Technology

Plan imetric control results of GPS surveying have been used widely in all kinds of engineering, while its height information is being researched at present. This paper introduces statistical learning theory and support vector machine, proposes a new method, support vector machine technology, to simulate quasi-geoid.Based on real GPS surveying datum, we did an experiment with support vector machine, neural network and polynomial technology to simulate quasi-geoid.The compared and analyzed test results show that the simulating accuracy of support vector machine achieved the same accuracy of neural network and polynomial. Moreover, support vector machine can settle many questions that neural network must face, such as real-time handling data, over-learning, convergence too slowly, falling into regional maximum easily etc.