Manufacturing Control Architectures

Robert W. Brennan

Central to the manufacturing goal of meeting the conflicting requirements of short production lead times and high quality standards is the issue of appropriate control architectures for manufacturing: i.e., what decision-making structures result in a modifiable, reliable, and fault-tolerant system? Experience has shown that traditional, centralised architectures can be quite inflexible to change and provide little fault tolerance. This has led industrial and academic researchers to the development of a spectrum of decentralised control architectures. At one end of the spectrum lie the hierarchical control architectures with theoretical foundations in organisational theory and large-scale system control theory. At the other end lie non-hierarchical, or "heterarchical", structures arising from more recent developments in distributed artificial intelligence and multi-agent systems. Since existing industrial and academic research on manufacturing control has focused on qualitative comparisons of alternative structures and has done little more than prove the concept of heterarchical control, work in this area has focused primarily on providing objective comparisons control architectures that span this spectrum.

In order to objectively compare different control architectures it is important to understand their individual structures more clearly. As a result, research in this area has focused on both the development of metrics to characterise and evaluate alternative control architectures as well as the development of experimental test beds.

Projects

• The identification of metrics for the analysis of alternative control architectures. Metrics include both structural parameters that are used to characterise given control architectures as well as manufacturing and control system performance measures.
• The development of test cases that can represent a common unbiased platform against which to compare and evaluate the performance of various proposed solutions. This common benchmark will give the research community a unique way to assess the opportunities and highlight the main pitfalls resulting from the adoption of MAS in the real industrial domain.

Selected Publications

R.W. Brennan and W. O, “Performance analysis of multi-agent scheduling and control systems for manufacturing,” Production Planning and Control, 2003. (accepted October 2002)

R.W. Brennan and D.H. Norrie, “Metrics for evaluating distributed manufacturing control systems,” Computers in Industry, 2003. (accepted September 2002)

R.W. Brennan, "Performance comparison and analysis of reactive and planning-based control architectures for manufacturing," Robotics and Computer Integrated Manufacturing, Vol. 16, No. 2-3, pp. 191-200, 2000.

Contact Information

        Email:          brennan@enme.ucalgary.ca
        Telephone:      (403) 220-5798

Last updated: 9 June 2003