基于神经网络的GPS高程拟合方法优选及精度分析

Optimization and Precision Evaluation with GPS Elevation Fitting Method Based on Neural Network

针对现有的几种神经网络GPS高程拟合方法,讨论了利用遗传算法(GA)、粒子群算法(PSO)优化BP神经网络权值和阀值的原理;结合分布较均匀、现势性较好的GPS和水准联测数据,试算了基于神经网络的GPS高程拟合。拟合结果表明:基于PSO算法优化的BP神经网络的拟合精度优于GA算法,误差相对更小。

According to current elevation fitting methods of networks,the genetic algorithms(GA) and particle swarm optimization(PSO) methods were employed to optimization of the weights and threshold of BP neural networks;with evenly distributed GPS data,GPS elevation fitting based on neural network is calculated.The fitting results show that optimization of the BP neutral network by PSO is better than that by GA and the error is relatively small.