Improving Water Security Analysis by Interfacing EPANET and CLIPS Expert Systems

In order to successfully manage critical infrastructure systems in particular those of water supply networks, there is a need to identify, assess and control various factors that may contribute to harming the supervised area. These factors have dependencies in a large set of heterogeneous information, which is often owned by different entities and data sources. All this information needs to be integrated to allow seamless use and consistent management, and after consolidation it can be reliably and quickly, communicated. This paper deals with the design of a Knowledge Based Expert System (KBES) able to provide an ‘on-line failures recognition and assessment' of tightly interconnected, technological infrastructures. This in turn contributes to a machine- readable, integrated, conceptual model of the whole system. In particular the aim of this paper is to present an intelligent software tool, using Artificial Intelligence (AI) techniques which allows the execution of an alarm analysis, through the remote sensing activity of complex plants. The AI component is based on CLIPS (C-Language Integrated Production System) which is a tool that offers a complete environment for developing a rule-based expert system. The AI component allows identification of the primary faults of the system, discriminating these former from other offshoot alarms. In other words, this tool shows which alarms are directly connected to primary faults and which alarms are consequential effects of the primary ones. The core of the software is an algorithm which uses a knowledge ontology and a set of alarm propagation rules which are both based on the Multilevel Flow Modelling (MFM) paradigm. The proposed algorithm has been tested integrating it with EPANET and by simulating some fault scenarios and analyzing the results.