| The problem of developing a dynamic scheduler, based on the principles of distributed control of manufacturing systems is considered in this research. The domain of this research is the scheduling of jobs in a shop floor environment to optimize a given regular or non-regular performance measure. In this environment, the arrival of jobs is not known a priori, jobs have several alternatives for processing, machines can be added or removed from the system and the machines are considered to be unreliable.; A generic mathematical model that accommodates an objective based on completion time is developed for the scheduling problem. The problem is decomposed into independent sub-problems based on jobs. These sub-problems are categorized into independent agents and they are used in the construction of an agent-based scheduler. The agents schedule jobs by optimizing their local objectives, which in turn are used to optimize the global objective in an iterative manner.; Shop floor dynamics such as machine failures are incorporated into the agent based scheduler by assuming probabilistic machine failure rates and Markovian process. The failure rate equations are used in determining the new time estimates to process a ready job on an idle unreliable machine.; The agent-based scheduler is tested using some known problems from the literature and randomly generated problems of varying dimensions. A duality gap of 4% to 17% is observed on problems with known solutions. When compared with the best solution obtained by using different dispatching rules an improvement of 20% to 32% is obtained. |
