A proposed structure for distributed shopfloor control

Abstract

Discusses the movement away from hierarchical organizational structures towards flatter, heterarchical, structures which is reflected in the growing interest in distributed manufacturing control systems. Traditional hierarchical control systems are limited by the breadth, quantity and timeliness of information needed for their operation. Distributed, heterarchical, control systems overcome these hierarchical limitations but, concurrently, forfeit advantages of the hierarchy including analytically optimal loading patterns and centralized pristine data tracking. Classifies existing research into four categories and documents a progression of heterarchical control approaches to inject some of the advantages of the traditional hierarchy into new heterarchical frameworks. Concludes that neither hierarchical nor heterarchical control structures are ideal in their pure form and, hence, proposes a modified structure, called the quasi‐heterarchical control system, which is a combination of, and a compromise between, pure hierarchy and pure heterarchy.

Application of reinforcement learning for agent-based production scheduling

“…Due to the structural rigidity of classical centralized control architectures in manufacturing, the decentralized (or heterarchical) control structure has drawn more attention (Crowe and Stahlman, 1995;Dilts et al, 1991;Duffie and Prabhu, 1994). One of the most important properties of the heterarchical structure is that the decision-making responsibilities are fully distributed to each component of the system.…”

Holonic Production System to Obtain Flexibility for Customer Satisfaction

“…Neither hierarchical or heterarchical systems are able to fulfil these requirements [3,4]. Hierarchical systems have a typically rigid structure which makes it very hard for them to react to turbulences in an agile way.…”

“…Holons and holarchies3 Most of the system architectures based on agents are heterarchical. Nevertheless there are also agent based systems which do not adopt heterarchical control 4. Holistic scientific paradigm focusing on the study of Complex Adaptive Systems (C.A.S.)…”

Iterative learning scheduling: a combination of optimization and dispatching rules

“…Panwalker and Iskander (1977) and Blackstone et al (1982) provided oft-quoted survey articles about known research results with sequencing rules. Studies have shown that the selection of the dispatching rules has a significant impact on system performance (Crowe and Stahlman, 1995). Given its importance, extensive research has been done in identifying dispatching rules that are superior.…”

“…The reason is that a dispatching rule ordinarily sequences jobs and/or machines only according to local information of the manufacturing system, either current system states or information of a specific manufacturing resource at the decision point. Consequently it cannot guarantee globally satisfactory performance (Brennan, 2000;Crowe and Stahlman, 1995).…”