Dependability in distributed computerized manufacturing systems

In order to achieve full dependability in computerized manufacturing systems there must be an automated fault detection and an automated reconfiguration of the system. However, for practical reasons there will often be enough to use a semi-automatic fault detection and reconfiguration, where a human operator can intervene in the system decisions. To be able to do automated reconfiguration the manufacturing system must be flexible. We will present a model suited for describing such manufacturing systems. Different reconfiguration methods will be described. Examples from a real system will be given. We will also give examples how automated fault detection can be achieved in manufacturing systems. This is influenced by the level of ambition that is chosen when designing the subsystems of the whole system. We will have to define a failure semantics for each subsystem. A third means for achieving dependability is to use buffering in the system. Here we need a strategy for the different subsystems, and also how to act when there is an emergency