欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
水质的时间变化趋势预测是进行水环境管理的前提，预测模型在很大程度上决定了预测精度的高低，如何选取有效的时间序列水质预测模型是目前的研究热点之一。以平西湖为研究对象，根据2009-2011年间TN、TP和CODMn月监测数据，提出了一种基于ARIMA和RBF-NN的组合模型，该模型能同时反映水质的渐变性和非线性变化的特点，最后用5个精度评价指标对组合模型的预测结果进行了评价，并和基于ARMMA和时间序列神经网络预测模型的预测结果进行了比较。结果表明，大部分指标显示ARIMA/RBF-NN组合模型对受内生变量影响较大的TN、TP的预测效果最好，ARIMA模型对受外生变量影响较大的CODMn的预测效果最优。
英文摘要：
Temporal trend prediction of water quality is the key to water environment management.How tochoose an effective prediction model of water quality time series is of intriguing interest at present.Inthis study,a combined model based on ARIMA and RBF-NN was developed,based on monthlymonitored TN,TP and CODMn between2009and2011from Pingxi Lake,which is one of the mainreservoirs in the Huaihe River basin.Water quality variation is a gradually changeable and nonlinearprocess,which is reflected in the model.The lags of the variable itself and its stochastic disturbance wereobtained from ARIMA model,and then were used as input variables of RBF-NN.The results indicatedthat prediction performance of the combined model ARIMA/RBF-NN was the best among ARIMA/RBF-NN,ARIMA and time series neural network for TN and TP both of which were mainly influencedby the endogenous variables.Nevertheless,the prediction performance of ARIMA was the best for CODMnwhich was greatly influenced by exogenous variables in this study area.
同期刊论文项目
同项目期刊论文
期刊信息
位置：