基于B样条神经网络的地磁场曲面拟合  

Based on the B-spline Neural Network of Geomagnetic Field Surface Fitting

作　　者：夏琦博 李建良[1] 

机构地区：[1]南京理工大学理学院,南京210094

出　　处：《安阳工学院学报》2017年第4期58-61,共4页Journal of Anyang Institute of Technology

基　　金：国家自然科学基金(41404053;41174165);江苏省高校自然科学研究基金(14KJB170012)

摘　　要：基于B样条神经网络对实测地磁数据进行曲面拟合,该网络不仅保留了B样条基函数在速度方面的改进,同时也继承了神经网络在计算精度方面的优势。选用了一种修正速度最快的改进的LMBP算法进行权值的修正,最后也通过Matlab仿真实验证明了基于B样条神经网络的拟合算法在速度与精度方面的优越之处。In this paper, based on the b-spline neural network to curved surface fitting of measured geomagnetic data, the network not only retained the b-spline basis function in the speed improvements, but also inherited the neural network advantage in terms of calculation accuracy.We choose the fastest a revised improved LMBP algorithm for correction of weights. Finally, through the Matlab simulation experiment proves the method based on b-spline neural network fitting method is superior in speed and precision.

关 键 词：B样条神经网络 曲面拟合 改进LMBP算法 

分 类 号：O241[理学—计算数学]