On-line industrial supervision and diagnosis, knowledge level description and experimental results C. Alonso Gonza Âlez a, * , B. Pulido Junquera a , G. Acosta Lazo b , C. Llamas Bello a a Grupo de Sistemas Inteligentes, Departamento de Informa Âtica, Universidad de Valladolid, Edi®cio de las Tecnologõ Âas de la Informacio Ân y las Telecomunicaciones, Camino del Cementerio s/n, E-47011 Valladolid, Spain b Grupo ADQDAT, Departamento de Electromeca Ânica, Facultad de Ingenierõ Âa, Universidad Nacional del Centro de la Provincia de Buenos Aires, Av. del Valle 5737, B7400JWI Olavarrõ Âa, Argentina Abstract This paper presents a detailed description of a knowledge-based system for on-line supervision and diagnosis of industrial continuous processes: TURBOLID. The system has been developed for a Spanish beet sugar factory, and it is in use in two plants. The system supports three main tasks; monitoring, operation mode and diagnosis. A detailed knowledge level account of the systems is presented in this paper, using CommonKADS methodology. The knowledge level description presented here will allow the reuse of the TURBOLID problem solving approach to supervision and diagnosis in other continuous plants, even in other domains. The main purpose of this paper is to illustrate how the basic tasks proposed are able to cope with supervision and diagnosis of a complex plant. Particularly signi®cant are the separation of operation mode and diagnosis, and the approach of TURBOLID to diagnosis, that obtains on-line causal explanations for detected problems. TURBOLID is able to differentiate among competing hypotheses looking for historical data, current data and even future data, tracking the plant evolution. Results of the activity of the system during a working month are also presented. q 2001 Elsevier Science Ltd. All rights reserved. Keywords: Supervision; Diagnosis; Knowledge-based system; CommonKADS experience model; On-line supervision of industrial processes 1. Introduction TURBOLID is a knowledge-based system (KBS), whose main objective is on-line plant-wide supervision of continuous industrial processes. It is one of the major outputs of a 4-year project `advanced process control and supervision through an expert system' that ended in 1997 and employed approximately twelve persons/year. It was a joint project between the former Sociedad General Azucarera de Espan Äa (S.G.A.E. S.A.), the University of Valladolid and the Spanish Open University. The main objective of the project was to improve the operation of a beet sugar factory through the introduction of advanced control and supervision techniques, mainly model-based predictive control and knowledge-based on-line super- vision of the plant. TURBOLID is a real-life system which evolved from the previous experiences of the research team, especially AEROLID, described in Alonso et al. (1998b). Conse- quently, it inherited some of its design principles; broad- ening its extent, to supervise more factory sections, and its scope, including more functionalities or widening the previous ones. When the project ®nished, a new version of the system, called TURBOGRAPH, was developed by company personnel. TURBOGRAPH is essentially the same system as TURBOLID, but with added graphical capabilities to monitor the KBS activity. Although TURBO- LID was designed for a particular plant, TURBOGRAPH is now in use in two factories. This work presents a detailed knowledge level descrip- tion of TURBOLID, using CommonKADS modeling methodology (Schreiber, Wielinga, & Breuker, 1993; Breuker and Van de Velde, 1994; Anjewierden and Schreiber, 1999). This provides a conceptual characteriza- tion of the system, independent of implementation details. The paper describes the experience model of TURBOLID, focussing on the task and inference layer because domain knowledge is more application-dependent. However, a domain ontology able to supervise TURBOLID basic tasks is under development. Particularly important is TURBOLID high-level task decomposition, based on a previously developed task taxonomy, introduced by Acosta (1995), and which, to some extent, may be applied Expert Systems with Applications 20 (2001) 117±132 PERGAMON Expert Systems with Applications 0957-4174/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved. PII: S0957-4174(00)00053-1 www.elsevier.com/locate/eswa * Corresponding author. Tel.: 134-983-423-000; fax: 134-983-423-671. E-mail address: calonso@infor.uva.es (C. Alonso Gonza Âlez).