论文部分内容阅读
本文简略地介绍了几种地质灾害数据处理与建模的非线性方法,主要包括GMDH自组织建模技术、神经网络方法。GMDH是一种高阶非线性回归建模方法,它是以简单的二元二次回归方程为基础,通过“代复一代”的“生产”过程,客观、自动地求得实际资料的非线性模型。而神经网络则是用工程技术手段模拟生物神经网络的结构特征和功能特征的一类人工系统。与常规统计方法相比,神经网络最突出的优点为它是通过对网络的学习和训练,来掌握变量之间的非线性关系。因此,其处理复杂问题的能力更强大。实例检验效果表明,这些非线性数据处理与建模技术考虑了地质灾害问题的非线性特性,其比基于常规统计理论的数据处理方法的精度要高得多。
This article briefly introduces several non-linear methods for data processing and modeling of geological disasters, including GMDH self-organizing modeling technology and neural network method. GMDH is a kind of high order non-linear regression modeling method, which is based on a simple binary quadratic regression equation, objectively and automatically obtained through the process of “generation by generation ” "production Nonlinear model of data. However, neural network is a kind of artificial system that simulates the structural and functional characteristics of biological neural network by means of engineering and technical means. Compared with the conventional statistical methods, the most prominent advantage of neural networks is that it grasps the nonlinear relationship between variables through the learning and training of the network. Therefore, its ability to handle complex issues is even stronger. The experimental results show that these non-linear data processing and modeling techniques take into account the nonlinear characteristics of geological hazard problems, which is much more accurate than the data processing methods based on conventional statistical theory.