| The last two decades have witnessed the emergence of new technologies in computer control and inter-computer communications. More recently, embedded controllers have redefined industrial automation through a new generation of “smart” machine tools (i.e., tools with local data processing capabilities).; New engineering technologies most often lead to new engineering approaches and, in turn, to new engineering practices. The focus of this dissertation is on the implications of these emerging technologies on manufacturing operations control.; Generally speaking, current industrial engineering practices continue to rely heavily on traditional methods for production control. These methods may be characterized by two common points: a unicentric view of the physical plant, and the treatment of the system components as passive elements, that is, uninvolved in the decision making process.; This study is a trial for a new paradigm, oriented more towards the discovery and evaluation of previously untested ideas. Specifically, it investigates the feasibility of distributed production line operations control where the information processing is locally carried out by networked (communicating) embedded station controllers. The decision making process relies on rule-based local learning from real-time process data. In accordance with the orientation of the study, the decision making process is kept intentionally simple to retain the focus on paradigm development rather than on seeking streamlined policies. Particularly, a minimal set of rules is selected to characterize the behavior of stations and a straightforward learning mechanism is chosen. The work includes a physical implementation and a computer simulation, whose findings are compared. The controller area network (CAN) is selected as the cost-effective communication protocol. A production line comprised of four stations is used as a testbed for the purposes of this study.; Evidence is revealed implying that even with simple rule-based learning, a distributed embedded approach achieves production line control performances comparable to the traditional techniques. Moreover, it is demonstrated that such a control system may be implemented using low-cost embedded control. The study discovers empirical aspects of distributed production control that present profound justification for future research. Perhaps most importantly, the distributed embedded approach shows promise in cases where system scalability requirements and environmental dynamics are more demanding. |
