An Agent Architecture for Supervising a Pilot WWTP

Contents

This paper describes the development and implementation of a real-time Knowledge-Based System (KBS) for the supervision and control of a wastewater treatment pilot plant (WWTPP) with biological removal of organic matter and nutrients. The hardware architecture contains different supervision levels, including two autonomous process computers (plant control and analyzers control) and a PLC, being the KBS the top supervisory level. The KBS has been developed using the G2 expert systems shell, and it is implemented for real-time operation. The implementation of the different components of the KBS is done under the Intelligent Agents paradigm. The knowledge is organized in several distributed agents, representing the available knowledge for every sub-process of the WWTPP. In addition to these independent agents, a Supervisor Agent acts as the master of the independent agents. The implementation of the KBS in the pilot plant has supposed the transformation of a classical control system with a fixed behavior in a system adaptable to different problems that could appear in a WWTP. The main achievement of this prototype is a versatile framework able to deal with different plant configurations, based on the object-oriented paradigm and on rule-based reasoning.