基于广义回归神经网络的趋势面自适应模型

Adaptive Model of Trend Surface Based on Generalized Regression Neural Network

趋势面分析方法是一种重要的物化探数据处理方法。构造趋势面常用的方法是进行多项式拟和,文章提出了一种利用广义回归神经网络趋势面建模方法,其中应用拟和度自适应调整神经网络参数。将该模型应用于某测区航放数据,处理结果表明了该模型不但拟和精度高,而且拟和度可调。

Analytical method using trend surface is a important data processing method in geophysical and geochemical techniques. The often used method of structuring trend surface is a polynomial equation. The article puts forward a method of structuring trend surface by using generalized regression neural network. It adaptively adjusts parameter of neural network according goodness of fit. The model is used to airborne radiocactive data of a surveyed area. The result indicates the model not only has high precision but also has adjustable goodness of fit.