A knowledge guided approach to the generation and analysis of manufacturing software requirements
Robert B. Rowen, Harvey G. Cragon, Gary W. Cobb · 1988
Manufacturing can be used as a strategic weapon and is an element in strategic business plans. Automation is a significant aspect of these strategies with the computer and its controlling software having key roles. Obtaining software requirements can be a major hurdle to success in planning a discrete manufacturing system's second level control software. This research addresses an observed impediment in the software requirements definition phase of an automation project. It is argued that deferred decision making and an inability of domain experts to articulate requirements makes prose requirements documents incomplete. An experienced software systems analyst can often guide design groups through possible choices and uncover areas of missing detail. An attempt to capture the analyst's experience requires an overall strategy and a specific criteria for completeness. This dissertation develops and demonstrates a methodology for using a queueing network model as a baseline for expert system augmented requirements definition. A rule based production system is used to guide the designer through potential choices concerning objects derived from the model. The responses to the questions are retained in man/machine readable files that can be post-processed to produce definition progress measures, prose descriptions and traceability of requirements. A general purpose modeling language is used for the queueing network models. The disassembly of this language, extendable to other general purpose languages, is documented. The methodology is directed toward discrete manufacturing processes. An electronic card manufacturing process is the primary technology used for examples.