Prediction of the bulking phenomenon in wastewater treatment plants L. Belanche a, * , J.J. Valde ´s a , J. Comas b , I.R. Roda b , M. Poch b a Seccio ´ d’Intel·lige `ncia Artificial., Departamento de Llenguatges i Sistemes Informa `tics, Universitat Polite `cnica de Catalunya, c/Jordi Girona 1-3, 08034 Barcelona, Spain b Laboratori d’Enginyeria Quı ´mica i Ambiental, Facultat de Cie `ncies, Universitat de Girona, Campus de Montilivi s/n, 17071 Girona, Spain Received 6 July 1999; revised 6 June 2000; accepted 6 June 2000 Abstract The control and prediction of wastewater treatment plants poses an important goal: to avoid breaking the environmental balance by always keeping the system in stable operating conditions. It is known that qualitative information — coming from microscopic examinations and subjective remarks — has a deep influence on the activated sludge process. In particular, on the total amount of effluent suspended solids, one of the measures of overall plant performance. The search for an input–output model of this variable and the prediction of sudden increases (bulking episodes) is thus a central concern to ensure the fulfillment of current discharge limitations. Unfortunately, the strong interrelation between variables, their heterogeneity and the very high amount of missing information makes the use of traditional techniques difficult, or even impossible. Through the combined use of several methods — rough set theory and artificial neural networks, mainly — reasonable prediction models are found, which also serve to show the different importance of variables and provide insight into the process dynamics.  2000 Elsevier Science Ltd. All rights reserved. Keywords: System identification; Wastewater treatment plants; Neural networks; Rough sets; Environmental modeling; Bulking 1. Introduction The world is an enormously complex system, where the environmental problems are essentially multi-faceted and demand at least a nodding acquaintance with many previously separated specialisms — ecology, economics, sociology, technology, physics, chemistry, and so on [1]. In particular, dirty water is both the world’s greatest killer and its biggest single pollution problem [2]. The large amount of wastewater generated in industrialized societies is one of the main environmental pollution aspects that must be seriously considered. New directives and regulations (for EU countries, the European directive of the Council 91/271/ EEC) have guaranteed the construction of specific plants to treat these wastewaters. Activated sludge systems are the most extensively used in wastewater treatment plants (WWTP). In an activated sludge process, the wastewater, which contains organic matter, suspended solids and nutri- ents, enters an aerated tank where it is mixed with biological floc particles. After a sufficient contact time, this mixture is discharged into a settler that separates the suspended biomass from the treated water. Most of the biomass is recirculated to the aeration tank, while a little amount is purged daily (see Fig. 1). Activated sludge is a clear example of an environmental process that is really difficult to understand, and thus diffi- cult to be correctly operated and controlled. The inflow is variable (both in quantity and in quality); not only is there a living catalyst (the microorganisms) but also a population that varies over time, both in quantity and in the relative number of species; the knowledge of the process is scarce, there are few and unreliable on-line analyzers, and most of the data related to the process are subjective and cannot be numerically quantified. It is known that most of the problems of poor activated sludge effluent quality result from the inability of the secondary settler to efficiently remove the suspended biomass from the treated water. When the biomass is strongly colonized by long filamentous bacteria, holding the flocs apart and hindering sludge settle- ment, the amount of Total suspended solids (TSS) at the outflow of the plant increases seriously. Although this phenomenon, called bulking, has been extensively studied, the interrelations and diversity of the many bacterial species Artificial Intelligence in Engineering 14 (2000) 307–317 0954-1810/00/$ - see front matter  2000 Elsevier Science Ltd. All rights reserved. PII: S0954-1810(00)00012-1 www.elsevier.com/locate/aieng * Corresponding author. Tel.: +34-93-401-5644; fax: +34-93-401-7014. E-mail addresses: belanche@lsi.upc.es (L. Belanche), valdes@lsi.upc.es (J.J. Valde ´s), quim@lequia1.udg.es (J. Comas), ignasi@lequia1.udg.es (I.R. Roda), manel@lequia1.udg.es (M. Poch).