Article Knowledge representation in an activated sludge plant diagnosis system George A. Vouros, 1 Ioannis S. Pantelakis 2 and Themistoklis D. Lekkas 2 (1) Department of Mathematics, University of the Aegean, Karlovassi, Samos, Greece (2) Department of Environmental Studies, University of the Aegean, Mytilene, Greece E-mail: {georgev,jpan,thl}aegean.gr Abstract: This paper focuses on the knowledge representation framework utilized by an integrated wastewater treatment expert system for diagnosing operational problems of an activated sludge plant. The system deals with events that may occur in all the units of an activated sludge plant and exploits on-line measurements, observations formed from data provided by laboratory analyses and empirical observations in an integrated manner. The system provides assistance to human experts to control the activated sludge process. It has been tested and evaluated in the pilot activated sludge plant of the Water and Air Quality Laboratory in the University of the Aegean. Keywords: diagnosis system, representation of uncertain and imprecise knowledge, causal graphs 1. Introduction The activated sludge (AS) process is the most common method of municipal wastewater treatment. The goal of the process is to produce a good effluent water quality (Tchobanoglous, 1991). However, many perturbations degrade the level of performance of an AS plant. Most of the perturbations are transient and will pass through a plant too quickly to be detected, but others are more persistent 226 Expert Systems, November 2000, Vol. 17, No. 5 and may reflect inappropriate control, causing the perform- ance of the AS plant to degenerate. The complexity of the overall AS process, the inadequate and unreliable on-line monitoring equipment, the absence of appropriate performance specifications on which to base control, the limited flexibility in plant design, the lack of accurate process models and the excess influent variability have shown that existing technology for diagnosing poten- tial operational problems in an AS plant and for controlling its performance has not been applied effectively (Barnett & Andrews, 1992). Whether the problem in an AS plant is transient or per- sistent, it is desirable to minimize the impact of effluent discharges in the environment. The environmental impacts of an upset can be significantly reduced by decreasing the time required to diagnose the problem and by performing the appropriate control actions. These actions aim to make the greatest improvement in plant performance in the short- est possible time. This paper describes an integrated wastewater treatment expert system for diagnosing potential problems of an AS plant. The system exploits on-line data, observations formed from data provided by laboratory analyses and empirical observations in order to provide assistance to human experts to control the AS process. The system has been tested and evaluated in the pilot AS plant of the Water and Air Quality Laboratory in the University of the Aegean. The paper focuses on the knowledge representation formal- ism utilized by the system and is structured as follows. Section 2 presents the wastewater plant supervision problem and refer- ences previous work done towards controlling AS plants. It describes the pilot plant in the Water and Air Quality Labora- tory of the University of the Aegean and presents the diagnosis and control tasks performed by human controllers at the knowledge level. Section 3 presents the knowledge represen- tation framework utilized, in relation to other advanced for- malizations for handling uncertain knowledge. Section 4 presents an actual case of using the implemented system. Finally, Section 5 concludes this paper and refers to issues that will be investigated in the future. 2. Wastewater plant supervision 2. 1. The problem According to Berthouex et al. (1985) 60% of perturbations are due to operational problems such as low dissolved oxygen (DO), low mixed liquor suspended solids (MLSS), solids handling and high flows. Furthermore, a literature review confirms that inadequate plant operation is respon- sible for the majority of wastewater treatment plant failures.