From AIRWiki
Jump to: navigation, search
Coordinator: FrancescoAmigoni (
Collaborator: AlessandroBellemo (
Research Area: warning.png"Agents, Multiagent Systems, Agencies" is not in the list of possible values (Affective Computing, Agents - Multiagent Systems - Agencies, BioSignal Analysis, Computational Intelligence and Games, Computer Vision and Image Analysis, E-Science, Machine Learning, Philosophy of Artificial Intelligence, Robotics, Social Software and Semantic Web) for this property.Agents, Multiagent Systems, Agencies
Research Topic: Multiagent Systems, Cooperative Negotiation
Start: 01/08/2012
End: 31/07/2013
Status: Active

Gestore dei Mercati Energetici S.p.A. (from here on “GME”) is the company that organizes and economically manages the Italian Energy Market. The Energy Market allows energy producers, consumers and wholesale customers to purchase and sell electricity, natural gas and green certificates.

Given the importance of energy in the modern days GME requires every company that operates on the Energy Market to provide financial guarantees; this way, if a company goes into default or cannot pay, the guarantee can be used instead; currently, two types of guarantee can be used when operating on the Energy Market: first-demand bank guarantees and non-interest-bearing cash deposits.

The problem with the current system is that different types of guarantee are required on different markets handled by GME; also the rules about how much guarantee is needed to cover operations vary from market to market. This makes it difficult for operators to distribute their available guarantee on all the markets they operate on and it also increases running costs for both operators and GME: operators have to get several different guarantees form their bank (each of which incurs a cost) and GME needs IT systems and employees to check whether the guarantee distributions provided by the operators are correct.

In order to overcome these problems GME decided to fund a research to create a prototype called FinAgent. FinAgent is a multiagent system that can automatically handle financial guarantees and distribute them on the markets on which an operator operates. In order to do this GME allows all guarantees provided by an operator to be used on every market so that guarantee distribution can be simplified. This also means that operators can reduce the number of guarantees they need and as such running costs are decreased. A multiagent system has been chosen because agents can create a model of the markets they handle and then try to foresee what kind of operations will be done on that market in the future based on the previous behaviour of the operator. In other words the guarantee distribution is based on: The rules of the market (e.g. whether operations must be fully or partially covered by guarantee). The past behaviour of the operator (e.g. if an operator often buys a lot of energy on a market on Monday then the agent in question will provide on Sunday a guarantee distribution that favours that market). Other miscellaneous rules (e.g. if many operations on a market have been cancelled due to lack of guarantee then in future the agent that handles that market will try to obtain more guarantee). These rules can also be set by the operator.

The current architecture of FinAgent is this: every operator has a stakeholder agent for each market it operates on and a moderator agent that coordinates all stakeholder agents. The guarantee distribution is obtained through cooperative negotiation.