The Advent
of Self-organizing Industrial Controls

The end of centralized, deterministic control systems

By : Jim Pinto,
San Diego, CA.
USA

This presentation was made at the Chaos in Manufacturing conference in Santa Fe, New Mexico, April '98.

The end is in sight for conventional, centralized control systems based on programmable controllers and DCS architectures - they simply cannot be scaled up. New peer-to-peer I/O based self-organizing controls are on the horizon.

This discussion was influenced, in part, by the book :

Look at Gleicks book CHAOS - Making a New Science By : James Glieck

This book (1987) re-introduced Chaos Theory and brought about the resurgence of interest in the new science of complexity.

Complete Presentation

The complete Powerpoint presentation is available for download.

ClickDownload a .pdf format file
Note : Approximate size : 560 kbytes.

Get the Adobe reader free Need a pdf reader? Go to the Adobe website

Outline

  • Industrial Control - Evolution or Revolution?
    • Yesterday - �Status Quo� - DCS/PLC Control Systems
    • Today - �Latest� developments - Networked I/O
    • Tomorrow - �Intelligent� I/O - Peer-to-peer �host-less� I/O
    • Future - �Self-organizing� Systems - Intelligent Agent-based Systems
  • Yesterday - Centralized Hierarchical control
    • DCS
    • PLCs
  • Today - Industrial Networks
    • Same paradigm - centralized controls
    • Fieldbus
    • BACnet
    • MAP
  • Disadvantages of Current Controls
    • Still largely Master-Slave architectures
    • PLC represents the control intelligence
    • DCS is the coordinating intelligence
    • Rooted in the Past
    • Deterministic algorithms
    • I/O is still relatively �slave� intelligence
    • Lots of different interface protocols
    • Struggle for inter-operability
  • Tomorrow is already here - Peer-to-peer I/O
    • �Autonomous� - self-acting I/O
    • Local intelligence (CPU and memory)
    • Peer-to-peer communications
    • No �Host� or Central Intelligence
    • The �Ghost-Host�
    • Central �supervisor� for configuration
    • Network computers provide HMI(MMI
  • What is the name for this new Constrol Science?
    • Chaos Control ?
    • Complexity Control ?
    • Autonomous I/O Systems ?
    • Peer-to-Peer Control Systems ?
  • Today - Autonomous I/O
    • Approaching �Intelligent Agents�
    • Peer-to-peer interactive controls
    • Algorithmic, rule-based controls
    • Can be made �robust� with algorithms
    • Fault-tolerance with redundancy
    • Still relatively �deterministic�
    • Programming rooted in past practice
  • Intelligent, Distributed Control
    • Networks of intelligent devices, called nodes, that communicate using a common protocol
    • Each node contains embedded intelligence that implements the protocol and performs the node's algorithmic control functions
    • Each node includes a media interface which couples the node's processing resources with the communications media
    • Peer-to-peer communication capability that does not require intervention by a master controller
  • �Holonic Fractal Manufacturing based on autonomous, multi-purpose, modular and re-configurable production lines/cells will allow production of many different products of any quantity�
    - Odo Struger - Allen-Bradley
  • Can intelligent systems become self-organizing? The essential ingredients
    • Re-programmability (by others) over the network
    • Ability to change �behavior� (self-re-programming) Based on external or internal stimuli/algorithms
  • Critical Complexity
    • When does processing power become intelligence?
    • When does connected intelligence become "self-organizing"?
  • Non-deterministic behavious of self-organized control systems
    • Can we make �control systems� self-organizing?
    • If we �give up�on deterministic behavior, we�d have to insert the equivalent of Asimov�s Laws of Robotics - Specifically dis-allow certain harmful characteristics, or behaviors
  • Pinto's Laws of Self-organizing Controls
    Control Systems will :
    • maximize benefits
    • minimize waste, and
    • prevent accidents or harmful effects
  • Optimum Complexity - Chaitin Corollary
    • The �task� of a self-organizing control system would be to achieve control using the most-effective (shortest) control-algorithm.
  • Stuart Kaufman - Santa Fe /95
    �Are there complex systems which cannot, in principle, be assembled by an evolutionary process?�
    Perhaps self-organizing control systems can �evolve� to levels of control and emergent complexity far beyond anything imagined by conventional �deterministic� control systems.
  • Stuart Kaufman
    • "When a system of simple inter-acting components reaches a certain level of complexity or inter-connectedness, it undergoes a dramatic transition, or phase change."
  • Chris Langton
    • A systems computational capability peaks in a narrow regime between highly periodic and chaotic behavior.
  • Who will make the revolutionary breakthrough?
    • �I am convinced that the nations and people who master the new sciences of complexity will become the economic, cultural and political super-powers of the next century�
      - Heinz Pagels - the Dreams of Reason ('87)

Web Hotlinks (click on any item)


Return to Index of all JimPinto Writings Return to Index of all JimPinto Writings
Return to Jimpinto.com Homepage Return to JimPinto.com HomePage

If you have ideas or suggestions to improve this site, contact: [email protected]
Copyright 2000 : Jim Pinto, San Diego, CA, USA