【摘要】 针对GPS高程与正常高程的转换问题,给出了基于径向基函数(Radial Basis Function,RBF)神经网络的模型。该模型隐含层选用Gauss函数作为基函数,学习算法采用自组织选取中心法策略。用实际观测数据进行了试验和仿真,结果表明,用RBF神经网络转换GPS高程的精度优于最小二乘法。对RBF神经网络方法和反向传播(Back-Propagation,BP)神经网络方法转换GPS高程的精度进行了比较,虽然两种方法的结果相近,但RBF神经网络方法在学习速度方面远比BP神经网络方法快。由此可见,RBF神经网络方法用于转换GPS高程是可行和有效的。更多还原

【Abstract】 Authors introduced a radial basis function(RBF) neural network model,which was applied to convert GPS height to normal height.Gauss function was chosen as the radial basis function in the hidden layer of the model,while the network learning algorithm adopted automatic clustering and least square method.Through the RBF neural network’s training and calculation of GPS testing data,the accuracy of RBF algorithm is better than that of the conicoid fitting method.RBF algorithm is much faster than the back-propagation(BP) algorithm,though they have the same accuracy by comparison.So the conversion of GPS height based on RBF neural network algorithm is feasible and effective.更多还原