论文部分内容阅读
利用人工神经网络(ANN),探讨在不无监测系统的集水区城市降水质量预测的适用性.预测使用常规的气候和地理数据集,通过构建背景传播的神经网络和回归联合模型,克服利用逐步回归的方法对数据进行分析时违背独立数据假设的问题.研究通过交叉验证用于确定停止降水时间为输入变量参数,利用地区平均浓度(EMC)作为独立的变量,构建的模型比用负荷量构建的模型更精确.数据域和输入变量的选择对回归模型的准确性也有较大影响.但计算效率、动量和隐节点数目的选择等因素,对人工神经网络模型准确性的影响较小.同时,回归和人工神经网络模型的降水质量预测结果十分相似,但在不无监测系统的集水区域城市降水质量的预测方面,回归模型更有实效性.
The artificial neural network (ANN) is used to explore the applicability of urban precipitation quality prediction in watersheds without monitoring system.Using conventional climatic and geographic data sets, this paper forecasts the use of background neural network and regression joint model Stepwise regression method to analyze the data violated the assumption of independent data.The study used crossover validation to determine the stopping precipitation time as the input variable parameters and the regional average concentration (EMC) as the independent variable, The accuracy of the regression model is greatly influenced by the selection of data fields and input variables, but the computational efficiency, the choice of momentum and the number of hidden nodes and other factors have little effect on the accuracy of the ANN model. At the same time, the regression results are similar to those of the artificial neural network model, but the regression model is more effective in predicting the urban precipitation quality in the catchment area without the monitoring system.