期刊名称：International Journal of Electrical and Computer Engineering
电子版ISSN：2088-8708
出版年度：2017
卷号：7
期号：3
页码：1620-1627
DOI：10.11591/ijece.v7i3.pp1620-1627
出版社：Institute of Advanced Engineering and Science (IAES)
摘要：The main objective of wastewater treatment plant is to release safe effluent not only to human health but also to the natural environment. An aerobic granular sludge technology is used for nutrient removal of wastewater treatment process using sequential batch reactor system. The nature of the process is highly complex and nonlinear makes the prediction of biological treatment is difficult to achieve. To study the nonlinear dynamic of aerobic granular sludge, high temperature real data at 40˚C were used to model sequential batch reactor using artificial neural network. In this work, the radial basis function neural network for modelling of nutrient removal process was studied. The network was optimized with self-organizing radial basis function neural network which adjusted the network structure size during learning phase. Performance of both network were evaluated and compared and the simulation results showed that the best prediction of the model was given by self-organizing radial basis function neural network.
其他摘要：The main objective of wastewater treatment plant is to release safe effluent not only to human health but also to the natural environment. An aerobic granular sludge technology is used for nutrient removal of wastewater treatment process using sequential batch reactor system. The nature of the process is highly complex and nonlinear makes the prediction of biological treatment is difficult to achieve. To study the nonlinear dynamic of aerobic granular sludge, high temperature real data at 40˚C were used to model sequential batch reactor using artificial neural network. In this work, the radial basis function neural network for modelling of nutrient removal process was studied. The network was optimized with self-organizing radial basis function neural network which adjusted the network structure size during learning phase. Performance of both network were evaluated and compared and the simulation results showed that the best prediction of the model was given by self-organizing radial basis function neural network.
