灰色神经网络在GPS高程拟合中的应用

Application of grey neural network in GPS height fitting

针对传统单一GPS高程拟合方法存在对数据和测区要求较高的问题,提出一种改进的灰色神经网络拟合方法:通过添加跳脱系数修正适应度函数扩大搜索范围,对果蝇优化算法进行改进,有效地提高算法的寻优能力;利用改进的果蝇算法优化灰色神经网络参数,训练灰色神经网络模型获得模型最佳参数,完成对GPS高程的拟合。实验结果表明该方法具有较高的拟合精度和较强的鲁棒性。

Aiming at the problem that the traditional single GPS height fitting method requires more demanding data and surveying area,the paper proposed an improved fitting grey neural network：function fitness was modified by adding the jump coefficients,and the algorithm of fruit fly was extended to improve the searching range,which could enhance the searching ability of the algorithm;the improved algorithm was used to optimize the parameters of grey neural network,and the optimal value was obtained by training the grey neural network model to fit the GPS height.Experimental result showed that the proposed method could have higher fitting accuracy and stronger robustness.