The P3 Formalism: A Basis for Improved Diagnosis in Complex Systems
Accident investigations of engineered systems show that abnormal conditions that propagate to accident status are a complex combination of factors related to plant, people and procedures. This we denote as the P3 formalism. The annual cost to Australian industry of inadequate abnormal condition management (ACM) can run into hundreds of millions of dollars or beyond. It remains a significant challenge to designers, operators and regulators.
In this paper we develop and illustrate a new functional systems framework for analysis of system faults using the P3 formalism. Structured analysis provides the logical development of real-time diagnostic tools for improved ACM. Development is aided via the adoption of formal information structures and extensible vocabularies that are represented in ontologies for the three component classes. This helps in significant reuse and exploitation of the knowledge typically generated within a project or retrofit.
Using the P3 formalism we develop complementary and “blended” hazard identification approaches, the outcomes of which form the basis for a computer-aided root cause analysis. This provides the knowledge base for construction of real-time diagnostic tools, which in this paper are illustrated through the development of a multi-agent system, implemented on a Protégé/Jess/JADE development platform. The outcome provides both pro-active and reactive diagnostic tools. A number of case studies help illustrate the framework and its application.
The work is a collaboration between The University of Queensland, BP Refinery, Bulwer Island and BlueScope Steel, Port Kembla. Academic partners include the Computer and Automation Research Institute and the University of Pannonia, Hungary.