Abstract :
Describes the application of an artificial intelligence planning system, SCINAPSE to a class of production scheduling problems. SCINAPSE is a domain-independent planner which operates on a model of some domain. Given objectives posed as goals within that domain, it can then create plans for their accomplishment. One such type of domain, which has been addressed is production scheduling, where the objectives are the output of a number of products and for whose manufacture a work schedule is required. Production facilities are modelled in terms of resources and process-plans for the product types being manufactured. Sets of work-orders can then be presented to SCINAPSE, as well as a definition of planned resource outages and shift times. SCINAPSE then constructs a production schedule to meet the required production output, perhaps relaxing some constraints such as due-dates if the original requirement is found to be unattainable. The current production scheduling prototype includes a graphical interface through which the model may be defined in a declarative fashion. A simple flexible manufacturing system (FMS) has been modelled and the paper refers to that example
