Abstracts
Résumé
La coagulation est l’une des étapes les plus importantes dans le traitement des eaux. La difficulté principale est de déterminer la dose optimale de coagulant à injecter en fonction des caractéristiques de l’eau brute. Un mauvais contrôle de ce procédé peut entraîner une augmentation importante des coûts de fonctionnement et le non-respect des objectifs de qualité en sortie de la station de traitement. Le sulfate d’aluminium (Al2SO4.18H2O) est le réactif coagulant le plus généralement utilisé. La détermination de la dose de coagulant se fait au moyen de l’essai dit de « Jar Test » conduit en laboratoire. Ce type d’approche a le désavantage d’avoir un temps de retard relativement long et ne permet donc pas un contrôle automatique du procédé de coagulation.
Le présent article décrit un modèle neuro flou de type Takagi Sugeno (TK), développé pour la prédiction de la dose de coagulant utilisée lors de la phase de clarification dans la station de traitement des eaux de Boudouaou qui alimente la ville d’Alger en eau potable. Le modèle ANFIS (système d’inférence flou à base de réseaux de neurones adaptatifs), qui combine les techniques floues et neuronales en formant un réseau à apprentissage supervisé, a été appliqué durant la phase de calage et testé en période de validation. Les résultats obtenus par le modèle ANFIS ont été comparés avec ceux obtenus avec un réseau de neurones de type perceptron multicouche (MLP) et un troisième modèle à base de regression linéaire multiple (MLR). Un coefficient de détermination (R2) de l’ordre de 0,92 en période de validation a été obtenu avec le modèle ANFIS, alors que pour le MLP, il est de l’ordre de 0,75, et que pour le modèle MLR, il ne dépasse pas 0,35. Les résultats obtenus sont d’une grande importance pour la gestion de l’installation.
Mots-clés :
- modélisation coagulant,
- réseaux de neurones,
- modèle neuro floue,
- analyse en composantes principales,
- régression linéaire multiple,
- station de traitement
Abstract
Coagulation is an important component of water treatment. Determining the optimal coagulant dosage is vital, as insufficient dosage will result in undesirable treated water quality. A number of chemicals have been used successfully in coagulation, particularly alum (Al2SO4•18H2O). Traditionally, jar tests are used to determine the optimum coagulant dose. However, this is expensive and time-consuming and does not enable responses to changes in raw water quality in real time. An optimal modeling approach can be used to overcome these limitations.
The main purpose of this study was to investigate the applicability and capability of Adaptive-Network-Based Fuzzy Inference System (ANFIS) and Neural Network (ANN) methods for modeling coagulant dose. To verify the application of this approach, Boudouaou surface water, located in the northern part of Algeria, was chosen as the case study area. The data used for the determination of the models included six (6) input variables describing the raw water characteristics (temperature, pH, turbidity, conductivity, dissolved oxygen and the ultraviolet absorption). The data set was divided into two (2) subgroups, calibration and validation periods. Coagulant models having various input structures were trained and tested to investigate the applicability of the used methods. To obtain a more accurate evaluation of the results for the ANFIS models, the best fit model structures were also tested by artificial neural network (ANN) and multiple linear regression (MLR) methods. The results of three methods were compared, and it was observed that the ANFIS is preferable and can be applied successfully because it provides high accuracy and reliability for coagulant dosage modelling, according to the following performance evaluation criteria: determination coefficient (R2), Root Mean Square Error (RMSE) and bias (B). The results are of practical importance: the coagulant dose changes according to the six variables representing the raw water characteristics, as there is no one dominant variable.
Keywords:
- coagulant modelling,
- neural networks,
- Fuzzy Inference System,
- principal components analysis,
- multiple linear regression,
- water treatment plant
Appendices
Références bibliographiques
-
ADGAR A., C.S. COX, P.R. DANIEL, A.J. Billington et A. Lowdon (1995). Experiences in the application of the artificial neural networks to water treatment plant management. Dans : Proceedings of the International COMDEM’95, Vol. 1, Canada, pp. 33-38.10.1139/cjce-28-S1-26 Google Scholar Search this bibliographic reference on Google Scholar
-
Adgar A., C.S. Cox et T.J. Böhme (2000). Performance improvements at surface water treatment works using ANN-based automation schemes. Transactions Inst. Chem. Eng., 78, Part A, 1026-1039.10.1016/0165-0114(94)90297-6 Google Scholar Search this bibliographic reference on Google Scholar
-
Amirtharajah A. et C.R. O’Melia (1990). Coagulation processes: Destabilization, mixing, and flocculation. Water quality and treatment. Pontius F.W. (Éditeur), McGraw-Hill, New York, NY, États-Unis, pp.269-365.10.1111/j.1747-6593.1990.tb01467.x Google Scholar Search this bibliographic reference on Google Scholar
-
Baxter C.W. (1998). Full-scale artificial neural network modelling of enhanced coagulation. Thèse de maîtrise, University of Alberta, Edmonton, Canada, 151 p.10.1016/S0273-1223(99)00641-1 Google Scholar Search this bibliographic reference on Google Scholar
-
Baxter C.W., S.J. Stanley et Q. Zhang (1999). Development of a full-scale artificial neural network model for the removal of natural organic matter by enhanced coagulation. J. Water Supp. Res. Technol. AQUA, 48, 129-136.10.1061/(ASCE)0733-9372(1991)117:3(308) Google Scholar Search this bibliographic reference on Google Scholar
-
Baxter C.W., Q. Zhang, S.J. Stanley, R. Shariff, R.R.T. Tupas et L. Stark (2001a). Drinking water quality and treatment: the use of artificial neural networks. Can. J. Civ. Eng., 28 (Suppl. S1), 26-35.10.1139/cjce-28-S1-26 Google Scholar Search this bibliographic reference on Google Scholar
-
Baxter C.W., R.R.T. Tupas, Q. Zhang, R. Shariff, S.J. Stanley, B.M. Coffey et K.G. Graff (2001b). Artificial intelligence systems for water treatment plant optimization. Am. Water Works Ass. Res. Found. et Am. Water Works Ass., Denver, CO, États-Unis, 141 p.10.1109/21.256541 Google Scholar Search this bibliographic reference on Google Scholar
-
Baxter C.W., R. Shariff, S.J. Stanley, D.W. Smith, Q. Zhang et E.D. Saumer (2002). Model-based advanced process control of coagulation. Water Sci. Technol., 45, 9-17.10.1623/hysj.2005.50.4.683 Google Scholar Search this bibliographic reference on Google Scholar
-
Bazer-Bachi A., E. Puech-Coste, R. Ben Aim et J.L. Probst (1990). Mathematical modelling of optimum coagulant dose in water treatment plant. Rev. Sci. Eau, 3, 377-397.10.1016/0043-1354(93)90044-I Google Scholar Search this bibliographic reference on Google Scholar
- BENKACI A.T. et N. DECHEMI (2004). Modélisation pluie débit journalière par des modèles conceptuels et « boîte noire »; test d’un modèle neuroflou. J. Sci. Hydrol., 49, 919-930.10.1016/S1364-8152(03)00163-4 Google Scholar Search this bibliographic reference on Google Scholar
-
Böhme T.J., C.S. Cox et A. LOWDON (1999).Performance assessment of a neuro self-tuning PI controller to be used at a water treatment plant. Dans : Proceedings of the American Control Conference, San Diego, CA, États-Unis, pp. 3216-3220.10.1007/BF02478259 Google Scholar Search this bibliographic reference on Google Scholar
-
Buckley J.J. et Y. Hayashi (1994). Fuzzy neural networks: a survey. Fuzzy Sets Sys., 66, 1-13.10.1016/0165-0114(94)90297-6 Google Scholar Search this bibliographic reference on Google Scholar
- CARDOT C. (1999). Les traitements de l’eau. Procédés physico-chimiques et biologiques. Ellipses Edition Marketing S.A., 247 p.10.1021/es030360x Google Scholar Search this bibliographic reference on Google Scholar
-
Cox C.S., L. Rietveld, A. Adgar et A. Van der Helm (2003). A new modelling and control system simulation environment for the rapid design of potable water systems. Dans : International Symposium on Advanced Control of Chemical Processes-Hong Kong ADCHEM. Hong Kong, http://www.ust.hk/adchem2003/.10.1016/S0019-9958(65)90241-X Google Scholar Search this bibliographic reference on Google Scholar
-
Critchley R.F., E.O. Smith et P. Pettit (1990). Automatic coagulation control at water treatment plants in the North-West region of England. Water Environ. J., 4, 535-543.10.1111/j.1747-6593.1990.tb01467.x Google Scholar Search this bibliographic reference on Google Scholar
- DECHEMI N., T. BENKACI et A. ISSOLAH (2003). Modélisation des débits mensuels par les modèles conceptuels et les systèmes neuro-flous. Rev. Sci. Eau, 16, 407-424.Google Scholar Search this bibliographic reference on Google Scholar
-
Dempsey B.A., H. Sheu, T.M. Tanzeer Ahmed et J. Mentink (1985). Polyaluminum chloride and alum coagulation of clay-fulvic acid suspensions. J. AWWA, 77, 74-80.10.1205/026387600528148 Google Scholar Search this bibliographic reference on Google Scholar
- DREYFUS G., M. SAMUELIDES, J. MARTINEZ, M. GORDON, F. BADRAN, S.THIRIA et L. HERAULT (2004). Réseaux de neurones - Méthodologies et applications. ÉDITIONS EYROLLES, Paris, France, 200 p. ISBN : 2-212-11464-8, www.editions-eyrolles.com.Google Scholar Search this bibliographic reference on Google Scholar
-
Edzwald J.K. et J.E. Tobiason (1999). Enhanced coagulation: US requirements and a broad review. Water Sci. Technol., 40, 67-70.10.1016/S0273-1223(99)00641-1 Google Scholar Search this bibliographic reference on Google Scholar
-
Edwards G.A. et A. Amirtharajah (1985). Removing color caused by humic acids. J. AWWA, 77, 50-57.10.1046/j.1365-2087.1999.00138.x Google Scholar Search this bibliographic reference on Google Scholar
-
Ellis G., A.g. Collins, X. Ge et C. Ford (1991). Chemical dosing of small water utilities using regression analysis. J. Environ. Eng., 117, 308-319.10.1061/(ASCE)0733-9372(1991)117:3(308) Google Scholar Search this bibliographic reference on Google Scholar
-
GAGNON C., B.P.A. GRANDJEAN et J.Thibault (1997). Modelling of coagulant dosage in a water treatment plant. Artif. Intel. Eng., 11, 401-404.10.2166/wst.2002.0539 Google Scholar Search this bibliographic reference on Google Scholar
-
Girou A., M. Franceschi, E. Puech-Costes et L. Humbert (1992). Modélisation des phénomènes de coagulation et étude de la morphologie des flocs : optimisation du taux de coagulant. Récents Progrès en Génie des Procédés, SFGP (Société Française de Génie des Procédés) (Éditeur), 6, 373-385, www.sfgp.asso.fr.
10.7202/705081ar
Google Scholar
Search this bibliographic reference on Google Scholar
- HEDDAM S., A. BERMAD et N. DECHEMI (2011). Applications of radial basis function and generalized regression neural networks for modelling of coagulant dosage in a drinking water treatment: A comparative study. J. Environ. Eng., Doi: 10.1061/ (ASCE) EE.1943-7870.0000435.Google Scholar Search this bibliographic reference on Google Scholar
- HORNIK K., M. STINCHCOMBE et H. WHITE (1989). Multilayer feedforward networks are universal approximators. Neural Networks, 2, 359-366.Google Scholar Search this bibliographic reference on Google Scholar
- HORNIK K., M. STINCHCOMBE et H. WHITE (1990).Universal approximation of an unknown mapping and its derivatives using multilayer feedforward networks. Neural Networks, 3, 551-560.Google Scholar Search this bibliographic reference on Google Scholar
- HORNIK K. (1991). Approximation capabilities of multilayer feedforward networks. Neural Networks ,4, 251-257.10.1016/0893-6080(91)90009-T Google Scholar Search this bibliographic reference on Google Scholar
-
Jang J.R.S. (1993). ANFIS: adaptive-network-based fuzzy inference system. IEEE Trans. Syst Man. Cybern., 23, 665-685.10.1109/21.256541 Google Scholar Search this bibliographic reference on Google Scholar
-
Jang J.R.S., C.T. Sun et E. Mizutani (1997). Neuro-fuzzy and soft computing. Prentice-Hall, Upper Saddle River, New Jersey, NJ, États-Unis, 614 p. www.prenticehall.com. ISBN-13: 978-0132610667.10.1002/j.1551-8833.1985.tb05512.x Google Scholar Search this bibliographic reference on Google Scholar
- JOLLIFFE I.T. (1986). Principal component analysis. Springer-Verlag, New York, NY, États-Unis, 519 pp.Google Scholar Search this bibliographic reference on Google Scholar
-
Kisi O. (2005). Suspended sediment estimation using neuro-fuzzy and neural network approaches. Hydrol Sci. J., 50, 683-696.10.1623/hysj.2005.50.4.683 Google Scholar Search this bibliographic reference on Google Scholar
-
Krasner S.W. et G. Amy (1995). Jar-test evaluations of enhanced coagulation. J AWWA, 87, 93-107.10.1016/S0954-1810(97)00010-1 Google Scholar Search this bibliographic reference on Google Scholar
-
LAMRINI B., M.V. LE LANN, A. BENHAMMOU et K. LAKHAL (2005). Detection of functional states by the “LAMDA” classification technique: application to a coagulation process in drinking water treatment. ElsevierC.R. Phys., 6, 1161-1168.Google Scholar Search this bibliographic reference on Google Scholar
- LEE C.C. (1990). Fuzzy logic in control systems: Fuzzy logic controller – part I and II. IEEE Trans. Sys., Man. Cybern., 20, 404-435.Google Scholar Search this bibliographic reference on Google Scholar
- LEFEBVRE E. et B. LEGUBE (1993). Coagulation floculation par le chlorure ferrique de quelques acides et phénols en solution aqueuse. Water Res., 27, 433-447.10.1016/0043-1354(93)90044-I Google Scholar Search this bibliographic reference on Google Scholar
- LEKFIR A., T. BENKACI et N. DECHEMI (2006).Quantification du transport solide par la technique floue, application au barrage de Beni Amrane (Algérie). Rev. Sci. Eau, 19, 247-257.10.1016/0893-6080(90)90005-6 Google Scholar Search this bibliographic reference on Google Scholar
- LIND C. (1994a).Coagulation control and optimization: Part one. Pub. Works, pp. 56-57, octobre.Google Scholar Search this bibliographic reference on Google Scholar
- LIND C. (1994b).Coagulation control and optimization: Part two. Pub. Works, pp. 32-33, novembre.10.1007/978-1-4757-1904-8 Google Scholar Search this bibliographic reference on Google Scholar
-
Maier H.R., N. Morgan et W.K. Christopher (2004). Use of artificial neural networks for predicting optimal alum doses and treated water quality parameters. Environ. Model. Software, 19, 485-494.10.1016/S1364-8152(03)00163-4 Google Scholar Search this bibliographic reference on Google Scholar
- MAMDANI E. (1977). Application of fuzzy logic to approximate reasoning using linguistic systems. Fuzzy Sets Sys., 26, 1182-1191.10.1016/j.crhy.2005.11.017 Google Scholar Search this bibliographic reference on Google Scholar
- McCULLOCH W. S. et W. PITTS (1943). A logical calculus of the ideas immanent in nervous activity. Bull. Math. Biophys., 5, 115-133.10.1007/BF02478259 Google Scholar Search this bibliographic reference on Google Scholar
- MIRSEPASSI A., B. CATHERS et H. DHARMAPPA (1997). Predicted of chemical dosage in water treatment plants using ANN and Box-Jenkings models. Dans : Preprints of 6th IAWQ Asia-Pacific Regional Conference. Korea, 16, pp. 561-1568.10.7202/013542ar
Google Scholar
Search this bibliographic reference on Google Scholar
-
Mohtadi m.f. et p.n. RAO (1973). Effect of temperature on flocculation of aqueous dispersions. Water Res., 7, 747-767.Google Scholar Search this bibliographic reference on Google Scholar
-
Nahm E., S. Lee, K. Woo, B. Lee et S.Shin (1996). Development of an optimum control software package for coagulant dosing process in water purification system. Dans : Proceedings of the Society of Instrument and Control Engineers Annual Conference, Tottori, Japon, 35, 1157-1161.Google Scholar Search this bibliographic reference on Google Scholar
- NAKOULA Y. (1997). Apprentissage des modèles linguistiques flous par jeu de règles pondérées. Thèse de Doctorat, Université de Savoie, France, 155 p.10.1109/TC.1977.1674779 Google Scholar Search this bibliographic reference on Google Scholar
-
Ratnaweera H. et H. Blom (1995).Optimisation of coagulant dosing control using real-time models selective to instrument errors. Water Supp., 13, 285-289.Google Scholar Search this bibliographic reference on Google Scholar
-
Rumelhart D.E., E. Hinton et J.Williams (1986). Learning internal representation by error propagation. Dans : Parallel Distributed Processing. Vol.1, MIT Press, Cambridge, Massachusetts, États-Unis, pp. 318-362.10.1016/0043-1354(73)90091-2 Google Scholar Search this bibliographic reference on Google Scholar
- SAPORTA G. (1990). Probabilités, analyse des données et statistique. Éditions Technip, Paris, France, 493 p.Google Scholar Search this bibliographic reference on Google Scholar
-
seaal. (2008). Société eau et assainissement d’Alger. Rapport interne. Google Scholar Search this bibliographic reference on Google Scholar
- SOUAG G.D., N. DECHEMI et A. BERMAD (2007).Simulation des débits mensuels en zone semi-aride par l’analyse en composantes principales et les modèles conceptuels. Sécheresse, 18, 97-105.Google Scholar Search this bibliographic reference on Google Scholar
-
Stum w., et j.j. MORGAN (1962). Chemical aspect of coagulation. J. AWWA, 54, 971- 992.10.21236/ADA164453 Google Scholar Search this bibliographic reference on Google Scholar
- TAKAGI T. et M. SUGENO (1985). Fuzzy identification of systems and its application to modeling and control. IEEE Trans. Sys. Man Cyber., 15, 16-132.Google Scholar Search this bibliographic reference on Google Scholar
-
Tutmez B., Z. HATIPOGLU et U. Kaymak (2006).Modelling electrical conductivity of groundwater using an adaptive neuro-fuzzy inference system. Comput. Geosci., 32, 421-433.Google Scholar Search this bibliographic reference on Google Scholar
-
Valentin n. (2000). Construction d’un capteur logiciel pour le contrôle automatique du procédé de coagulation en traitement d’eau potable. Thèse de doctorat, UTC/Lyonnaise des Eaux/CNRS, 153 p.Google Scholar Search this bibliographic reference on Google Scholar
-
Valentin N., T. Denoeux et F. FOTOOHI (1999). An hybrid neural network based system for optimization of coagulation doing in a water treatment plant. Dans : Proceedings of IJCNN99, Washington DC., IEEE, pp. 3380-3385.10.1002/j.1551-8833.1962.tb00922.x Google Scholar Search this bibliographic reference on Google Scholar
-
van Leeuwen J., C.W. Chow, D. Bursill et M. Drikas (1999). Empirical mathematical models and artificial neural networks for determination of alum doses of southern Australian surface waters. J. Water Sci. Res. Technol. Aqua, 48, 115-127.10.1109/TSMC.1985.6313399 Google Scholar Search this bibliographic reference on Google Scholar
-
Wang L. et J.M. Mendel (1992A). Fuzzy basis functions, universal approximation, and orthogonal least squares. IEEE Trans. Neural Networks, 3, 807-814.10.1109/72.159070 Google Scholar Search this bibliographic reference on Google Scholar
-
Wang L., et J.M. Mendel (1992b). Back-propagation fuzzy system as nonlinear dynamic system identifiers. Dans : Proc. of the IEEE Int. Conf. on Fuzzy Systems, San Diego, CA, États-Unis, pp. 1409-1416.Google Scholar Search this bibliographic reference on Google Scholar
-
Weishaar J.L., G.R. Aiken, B.A. Bergamaschi, M.S. Fram, R. Fujii et K. Mopper (2003). Evaluation of specific ultraviolet absorbance as an indicator of the chemical composition and reactivity of dissolved organic carbon. Environ. Sci. Technol., 37, 4702-4708.10.1021/es030360x Google Scholar Search this bibliographic reference on Google Scholar
- YU R., S. KANG, S. LIAW et M. CHEN (2000). Application of artificial neural network to control the coagulant dosing in water treatment plant. Water Sci. Technol., 42, 403-408.10.1046/j.1365-2087.1999.00135.x Google Scholar Search this bibliographic reference on Google Scholar
- ZADEH L. (1965). Fuzzy sets. Inform. Control, 8, 338-353.10.1016/S0019-9958(65)90241-X Google Scholar Search this bibliographic reference on Google Scholar
- ZADEH L. (1971). Quantitative fuzzy semantics. Inf. Sci., 3, 159-176. 10.1016/S0020-0255(71)80004-X Google Scholar Search this bibliographic reference on Google Scholar