EXPERT-MM: A knowledge-based system for maintenance management

This paper presents an approach to the design and development of knowledge-based systems in general and their application in the field of maintenance management in particular. The present approach is based on the idea that different kinds of knowledge in a given domain, namely declarative, procedural and heuristic are supported by corresponding methods and software tools. A prototype knowledge-based system, called EXPERT-MM, for the maintenance activities in the Siam Gipsum Industry (Bangkok, Thailand) has been worked out as a case study and is described in the paper. EXPERT-MM supports three main functions: maintenance policy suggestions, machine diagnosis and maintenance scheduling. The maintenance policy deals with the three types of preventive amintenance. For each component of the equipment it analyses the historical failure data and recommends an appropriate policy with optimal preventive maintenance intervals. This is based on the experts’ knowledge stored in a knowledge base. A rotary screw type air compressor is selected for a diagnosis. The knowledge representation scheme is rule-based and the inference strategy mechanism is backward chaining. The knowledge-acquisition process has been organized and realized using a decision tree diagram. The knowledge base contains 154 rules for the diagnosis and 54 rules for the maintenance model selection. The maintenance scheduling module is procedure based. EXPERT-MM development is based on the software tools dBase III Plus, TURBO PASCAL version 6.0 and expert system shell EXSYS, all integrated into a single software system with a user-friendly interface.