摘要
该文以MATLAB6 5为平台介绍了神经网络工具箱中径向基函数网络的基本原理、训练算法以及实现函数,指出该实现函数具有自适应确定网络结构和无需人为确定初始权值的特性。将其应用于某市的地表水质评价,研究了训练样本集、检测样本集及其目标输出的构造、原始数据的预处理、神经网络的构建、训练、检测及结果评价整个过程,收到了良好效果,为地表水质评价中神经网络的应用与设计奠定了扎实的基础。另外,还与BP网进行了对比,BP网表现出结构和初始权值确定的人为性。
On the basis of introducing the principles of RBF network, training methods and realizing functions in the ToolBox of MATLAB 6.5, this paper points out that the realizing function has advantageous properties such as adaptation for determining the network construction and independence of the output on initial weight value. A favorable outcome appeared after we apply this function to evaluating the quality of the surface water environment in a city, studying a whole process in the construction of training sample assemble,checking sample assemble and output targets , pretreatment of original data, establishment, training, inspection and result evaluation of the neural network. It lays a solid foundation for the application and design of neural network in surface water environmental quality assessment. On the other hand, comparing RBF net and BP net, BP net has random of determining the network construction and initial weight value.
出处
《地理与地理信息科学》
CSSCI
CSCD
北大核心
2003年第5期77-81,共5页
Geography and Geo-Information Science
基金
国家自然科学基金专项基金资助项目(40242018)
关键词
地表水
水质评价
径向基神经网络
人工神经网络
BP网络
surface water
assessment of water quality
RBF network
BP network
artificial neural network