优化NARX神经网络对时间序列溶解氧的预测

为提高非线性有源自回归(NRAX)神经网络模型的预测精准度,采用主成分分析(PCA)法和灰色关联分析(GRA)法提取原始数据特征并减少输入变量的维度,通过构建PCA-NARX和GRA-NARX模型预测地表水体未来短期(48 h)溶解氧(DO)的质量浓度.结果表明:GRA-NARX模型对时间序列DO质量浓度的预测效果优于...     

电子商务物流配送体系考核指标的建立与应用

配送体系的绩效考核是配送系统的重要内容,要建立和完善适应电子商务的物流配送体系,必须研究如何建立合理的配送指标体系。文章针对上海英业达公司的实际,探讨建立了一套绩效评估指标体系,并利用层次分析法对该配送体系进行了绩效评价。     

基于优化非线性自回归神经网络模型的水质预测

针对突发性水污染事件频发的问题,以上海市某支流具有代表性的监测断面为研究对象,通过优化调整输入数据段以及延迟阶数与隐含层神经元数等模型参数,构建基于历史水质时间序列的优化非线性自回归(NAR)神经网络模型,预测分析pH、溶解氧(DO)质量浓度和浊度3项水质指标的变化趋势。结果表明：优化后的NAR神经网络模型具有较好的非线性处理能力;当输入数据量为180,pH、DO质量浓度和浊度的神经网络模型的延迟阶数分别为2、3、9,隐含层神经元数为10时,NAR神经网络模型对pH、DO质量浓度和浊度的预测均方根误差分别为0.053、0.382 mg/L和17.300 NTU,平均绝对百分比误差分别为0.53%、3.97%和18.01%,预测效果较好。     

Removing Iron and Manganese Simultaneously from Ground Water Using One-stage Biological Filter

A novel process for removing iron and manganese simultaneously in ground water,which consisted of simple aeration and one-stage filtration,was developed in this research. It was found that the biological process had much higher manganese removal efficiency than chemical contact oxidation process. At the same time,the optimal operation parameters of aeration and biological filtration such as DO concentration and pH after aeration,filtration rate before and after startup,filtration operation cycle and backwas...     

Adsorption of Phenol by Activated Carbon Prepared from Waste Rice Straw by Chemical Activation with （ NH4 ） 2 HPO4

Activated carbons,RS-1,RS-2,and RS-3,prepared from rice straw with (NH4 )2HPO4 activation were used as adsorbents for removing phenols from aqueous solutions. The maximum adsorption capadties of phenol...