摘要
利用层次分析法构建符合区域中小河流健康评价指标体系和分级标准,基于RBF与GRNN神经网络算法原理,分别构建RBF与GRNN神经网络算法的河流健康评价模型,采用内插法构造网络训练样本,将河流健康分级评价标准值作为"预测"样本进行"预测",并将结果作为河流健康等级评价的划分依据,对文山州区域中小河流健康状况进行评价分析。结果表明:①RBF与GRNN神经网络模型对区域中小河流健康评价结果完全相同,与BP神经网络评价结果基本相同,表明研究建立的河流健康评价模型和评价方法均是合理可行的,同BP网络算法相比,RBF与GRNN神经网络模型有收敛速度快、预测精度高、不易陷入局部极小值等优点,且调整参数较少,只有一个SPREAD参数,可以更快地预测评价网络,具有较大的计算优势。②文山州区域主要河流健康评价等级为Ⅱ~Ⅲ级,即处于健康与亚健康之间,客观反映了区域中小河流健康状况,可为区域河流的可持续管理和生态环境建设提供参考依据。
The use of AHP to build medium-sized and small rivers in line with regional health evaluation system and grading stand- ards, based on the RBF and GRNN neural network algorithm theory, RBF and GRNN neural network algorithm river health assess- ment model is established, interpolation network training samples are constructed, the classification of river health assessment stand- ard value as "anticipate," sample "forecast", and as a result of the division of river health assessment based on grades, medium-sized and small rivers in the region of health evaluation and analysis in Wensharu The results show that: @RBF and GRNN neural net- work model of regional small river health assessment results are identical, the results of evaluation with BP neural network are basi- cally identical, indicating that the research on establishing a model for river health assessment and evaluation methods is reasonable and feasible, compared with BP network algorithm ratio, RBF and GRNN neural network model with fast convergence, the predic- tion accuracy is high, not easy to fall into local minimum, etc. , and adjust few parameters, only one SPREAD parameters, predic- tion and evaluation of the network can be faster, with greater computing advantage. QWenshan region's main river health evaluation Ⅱ-Ⅲ grade level, that is between health and sub-health, an objective reflection of the health of small and medium-sized rivers. This paper serves as a reference for the sustainable management of the regional rivers and ecological environment construction.
出处
《中国农村水利水电》
北大核心
2012年第3期56-61,共6页
China Rural Water and Hydropower