Use of hybrid models in wastewater systems

被引:29
作者
Anderson, JS
McAvoy, TJ [1 ]
Hao, OJ
机构
[1] Univ Maryland, Dept Chem Engn, Syst Res Inst, College Pk, MD 20742 USA
[2] Univ Maryland, Dept Civil Engn, College Pk, MD 20742 USA
关键词
D O I
10.1021/ie990557r
中图分类号
TQ [化学工业];
学科分类号
0817 ;
摘要
Incomplete first-principles knowledge of a system can obstruct the development of models for investigating the system's dynamic behavior. One approach proposed to overcome this problem involves using hybrid models. Hybrid models build inasmuch prior knowledge as available and then use empirical components, such as neural networks, to complete the description. In this paper hybrid modeling techniques are applied to two wastewater systems. The first system involves data from a Danish wastewater plant, and the second involves using a hybrid model for control. Potential problems involving the use of hybrid models are discussed.
引用
收藏
页码:1694 / 1704
页数:11
相关论文
共 21 条
[1]   DYNAMIC MODELING OF THE ACTIVATED-SLUDGE PROCESS - IMPROVING PREDICTION USING NEURAL NETWORKS [J].
COTE, M ;
GRANDJEAN, BPA ;
LESSARD, P ;
THIBAULT, J .
WATER RESEARCH, 1995, 29 (04) :995-1004
[2]  
CUTLER C, 1979, AICHE NAT M HOUST AP
[3]  
Cybenko G., 1989, Mathematics of Control, Signals, and Systems, V2, P303, DOI 10.1007/BF02551274
[4]  
Dors M., 1995, 6 INT C COMP APPL BI, P72
[5]  
GUJER W, 1995, WATER SCI TECHNOL, V31, P1, DOI 10.1016/0273-1223(95)00175-M
[6]   Alternating aerobic-anoxic process for nitrogen removal: Process evaluation [J].
Hao, OJ ;
Huang, J .
WATER ENVIRONMENT RESEARCH, 1996, 68 (01) :83-93
[7]  
Henze M., 1986, 1 IAWPRC
[8]   MULTILAYER FEEDFORWARD NETWORKS ARE UNIVERSAL APPROXIMATORS [J].
HORNIK, K ;
STINCHCOMBE, M ;
WHITE, H .
NEURAL NETWORKS, 1989, 2 (05) :359-366
[9]  
MCAVOY TJ, 1997, 14 WORLD C INT FED A
[10]   Application of artificial neural networks for crossflow microfiltration modelling: ''Black-box'' and semi-physical approaches [J].
Piron, E ;
Latrille, E ;
Rene, F .
COMPUTERS & CHEMICAL ENGINEERING, 1997, 21 (09) :1021-1030