Microeconomic Distributed Control: Theory and Application of MultiAgent Electronic Markets Hans Akke

1
Microeconomic Distributed ControlTheory and
Application ofMulti-Agent Electronic
MarketsHans Akkermans1,2, Jos Schreinemakers2,
Koen Kok3
1EnerSearch AB, Sweden2Free University
Amsterdam, The Netherlands3Energy research
Centre of the Netherlands (ECN)
2
Contents

• Setting the Scene
• Multi-Agent Systems in Energy
• Market-based Control
• How do Electronic Markets Work?
• Market-based Control
• Application Supply-Demand Matching
• Current Investigations
• Achievements and Issues

3
Development Trends in Distributed Intelligence

• Step 1 Internet plus World Wide Web
• To be considered as universally available
  baseline infrastructure
• Step 2 Pervasive computing
• Each and every device will have built-in
  computing power
• Computer itself becomes invisible like
  electromotor
• Step 3 Ambient intelligence
• Local intelligence plus global communication
• Semantic Web next-generation web, making the
  step from data handling to working with meanings
• Intelligent Agents computers and other devices
  can talk to, negotiate, make decisions, and
  cooperate with each other, and with humans.
• Step 4 IS transform into Information
  ecoSystems
• Info organizations of collaborating humans and
  systems
• Intelligence in IS systems that are aware of
  user context

4
MAS Energy Smart Buildings

• Intelligent Internet solutions for building
  control and energy management, realized by
  agent-mediated e-markets
• SMART project field experiment in office
  building (ECN, Petten, NL), run May-June 2002,
  including technical and social science studies
• Simple voting procedure for user interaction
• New field experiments underway (IIGO, Kropman BV,
  Nijmegen, NL)

5
How do Electronic Markets Work?
6
How do Electronic Markets Work?
7
Distributed Control Agents

• Xerox Parc office building California
• Climate control of rooms cooling during summer
• Cool air is scarce
• Local interest versus global optimum

8
Control Engineering Microecomonics Unified
9
Control Engineering Microecomonics Unified
10
Control Agents Scarce Resources
11
Microeconomic Distributed Control

• It can be shown that
• eMarkets handle scarce resources optimally both
  locally as well as globally (societally).
• If scarcity is absent operation as collection of
  local controllers.
• The market solution an omniscent central
  controller have identical outcomes.

Local data Market communication Central
control
12
Supply Demand MatchingCrisp Field Test
13
Supply Demand MatchingResidential Area with DG
14
Current InvestigationsIndustrial Applications

• Decentral load management
• Peak shaving
• On-line demand response
• Demand-supply matching in liberalised business
  settings
• Virtual Power Plant concepts
• Commercial portfolio management
• System support services delivery
• Agent-based comfort and energy management in
  large buildings

15
Achievements and Issues

• Achievements
• Agents do provide new solutions and services
• Design can be more flexible, open, robust
• Easy integration of domain with economic
  considerations
• Conceptually attractive and understandable

• Issues
• Agent society level theory
• Market protocols
• Customer value representations
• User-technology interaction
• e-Business models and services
• Empirical field studies
• IT research methodology