Supporting fault diagnosis through a multi-agent architecture

Abstract

In the industrial environment a short down-time of computer numerical control (CNC) machine tools is increasingly important. One approach to counter the loss of production is to make the machines more reliable, thereby reducing their times of non-use. Another is to enable machine operators to locate and to remove machine faults and their causes themselves, in order to reduce delay until maintenance staff is sent from a location far away. Operators may be supported in the fault-finding process by a knowledge-based decision-support system. The questions as to how to avoid the negative consequences of using such systems and how to even increase competence on the job is addressed by ComPASS, a system based on a multi-agent architecture. A running, functioning prototype was developed and evaluated. Based on this experience the system is currently being extended and improved, first results of which are presented here.