基于改进BP神经网络的GPS高程拟合方法

BP神经网络用于GPS高程拟合时存在收敛速度慢,受初始值选取影响大和易陷入局部极大值的问题。本文提出一种改进的BP神经网络高程拟合方法,将模拟退火算法(Simulated Annealing,SA)引入BP神经网络模型,利用模拟退火算法的全局寻优能力对BP神经网络的初始值进行选择,同时优化神经网络的各层神经元之间的连接权值和阈值,提高BP神经网络拟合法的拟合精度、收敛速度和推广泛化能力。最后结合实际算例对所提方法的拟合性能进行验证,结果表明利用模拟退火算法改进的BP神经网络进行高程拟合是可行且有效的,拟合结果优于传统BP神经网络法。

GPS Height Fitting Method Based on Improved BP Neural Network

BP neural network is used for GPS elevation fitting,which has a slow convergence speed,greatly affected by the initial selection and easy to fall into the local maximum value.In this paper,an improved BP neural network height fitting method for is proposed.The simulated annealing(SA)algorithm is introduced into the BP neural network model.The global optimization capability of the SA algorithm is used to optimize the connection weights and thresholds of the BP neural network neurons.The accuracy,convergence speed and generalization ability of BP neural network are improved.Finally,the performance of the proposed method is verified by a practical example.The results show that it is feasible and effective to use the BP neural network modified by simulated annealing algorithm.The result of the fitting is due to the traditional BP neural network method.