摘要
给水管网水质是多种污染因子的综合作用结果。为克服水质综合评价过程中的随机性与评价专家主观上的不确定性,利用神经网络的非线性和良好的函数逼近特性,建立了给水管网水质评价的广义回归神经网络(GRNN)模型。该模型具有网络结构自适应确定及输出与初始权值无关等优良特性。通过对在水质评价各等级间随机内插足够数量的训练样本的训练,确定合适的光滑因子。通过实例验证了模型评价结果与实际情况的一致性,为给水管网水质的评价提供了一种新方法.
Water quality in water supply networks is affected by a number of factors.In order to avoid the random in the process of evaluation and idealistic uncertainty of experts,an evaluation model for water quality in water supply networks is constructed by using a general regression neural network(GRNN),which is nonlinear and has excellent character of function approximation.The GRNN model with a adaptive network structure and other excellent characteristics,the output has nothing to do with the initial weights.Enough training samples are produced through interpolation between grades of the water quality evaluation and the smoothing factor is determined.Example suggests good agreements between the result of evaluation and actual measuring data and it presents a new method for evaluation of water quality in water supply networks..
出处
《沈阳理工大学学报》
CAS
2011年第4期63-66,共4页
Journal of Shenyang Ligong University
关键词
给水管网
广义回归神经网络
综合评价
water supply networks
general regression neural network
synthetic evaluation