核心期刊网首页> 外文会议> 机械 & 仪表工业

Application of neural networks to predictive modelling of the separation process by pervaporation


For the very first time artificial neural networks (ANN) have been used as a novel but reliable approach for the modelling and prediction of the pervaporation separation performance of various aqueous ethanol solutions by a process using composite membranes. The concentrated ethanol may be used, in place of petrol or liquefied petroleum gas, to power internal combustion engines in transport and power industries. Two neural network models have been developed to predict the typical separation performance. These models consisted of the feed forward multilayer(s) neural network. Inputs to the neural networks comprised of operating variables such as: temperature, feed composition, weight percent of the blended ingredients. Other novel parameters were extracted from Fourier Transformed Infra Red/ Attenuation total reflection (FT IR/ATR) spectra of each composite membrane. Water and ethanol permeation rates were the output from two neural net models. Very good correlation coefficients were obtained when tested with the set of test data. The correlation coefficients were found to be better than 0.9 with r.m.s. error of less than 0.069.......

【作者名称】: S Walpalage, V N Malhotra, A R Mirzai
【关 键 词】: Application of neural networks to predictive modelling of the separation process by pervaporation
【会议名称】: Sixteenth National Conference on Manufacturing Research University of the East London, UK 5-7 September 2000
【期刊论文数据库】: [DBS_Articles_01]
【期刊论文编号】: 101,087,007
【摘要长度】: 1,145
【会议地点】: London(GB)
【会议时间】: 2000
【上篇论文】: 外文会议 - DESIGN OF RECONFIGURABLE MANUFACTURING SYSTEMS WITH STRONGLY COUPLED NATURE
【下篇论文】: 外文会议 - Large ultralightweight optic fabrication: a manufacturing technology for advanced optical requirements

【论文下载】: 免费获取 该期刊&论文全文内容