An object-oriented, distributed approach to the development of integrated manufacturing systems.

Abstract

The total effectiveness of a manufacturing enterprise relies heavily on the degree of integration of all systems in the enterprise. However, manufacturing systems integration is difficult due to the extreme complexity, continual changes, and time constraints involved in manufacturing, and because most manufacturing practice is experience-based. This dissertation proposes an object-oriented, distributed approach to help designers in the development of integrated systems. The approach includes a systems modeling and design method, a framework for integrated systems, an example application, and a simulation. The modeling and design method is object-oriented and explicitly supports abstraction, encapsulation, classification, and generalization of manufacturing components. This method encourages knowledge accumulation and reuse. The method is particularly suitable for the development of very-large integrated systems because it has a higher level of abstraction than most of existing object-oriented methods and because it separates intra-object and inter-object integration. Another distinct feature of the method is the ability for rigorous description and mathematical analysis of the collaboration relationship and asynchronous communication between objects. The framework specifies three structures for uniformly modeling all objects in an integrated manufacturing system in terms of their generalization, composition, and collaboration relationships. The most innovative feature of the framework is the collaboration model which unifies the supervisory and cooperative control/command relationships in a manufacturing system. An integrated system which is developed based on this framework is more distributed, and more flexible and available than if based on existing frameworks. The feasibility of this approach is demonstrated by the design and simulation of an integrated shop-floor production system. The simulation also provides insight into the relationship between the control/communication and production systems. The simulation methodology used incorporates more sophisticated statistical methods (i.e. experimental design and multiple regression) than those used in most existing simulation studies of.