Reference: GV04A/543
Period underway: 01/01/2004 – 01/01/2006
Type: Research projects
Amount: 14,186 €
Financed by: CONSELLERIA DE EMPRESA, UNIVERSITAT I CIENCIA
Principal Investigator: García Sabater, José Pedro
Participant: Andres Romano, Carlos; Cardos Carboneras, Manuel; Garcia Sabater, Jose Pedro; Gomez Gasquet, Pedro; Miralles Insa, Cristobal Javier

Description:

The project objective is to face the real complexity of the production scheduling problem in the supply chain environment in the ceramic sector. Traditionally, the production scheduling problem has been modelled in an attempt to simplify as much as possible the conditions in the environment, and to improve, in most cases, a single performance parameter. However, this project takes us closer to a much more complex reality from both the environment point of view (requirements and restrictions) and the objectives point of view.

The environment intended to be modelled in this project must reflect the dynamic conditions of constant change that correspond to the ceramic supply chain; e.g. dynamic entry of jobs, non-availability of machines through breakdown, sequence-dependent delivery times (paving and coating factory), close locations of the various chain components, variation in components, relationships and the chain’s requirements, etc.

Regarding the objectives, the project will take into account the combination of objectives/interests which, at various levels (supply chain, companies, other processes-planning, purchases, sales, etc.), are affected by the decisions made during the Production scheduling process. In the project a scheduling method will be proposed, based on an agreement reached by the parties, to achieve overall company profits. Performance measures raised from an exclusive scheduler viewpoint (average flow time, deviation from the delivery date, etc.), may not benefit the general interest, especially in a changing environment.

To meet the project purposes, the use of Multi-agent Systems is proposed. Agent-based software paradigm development proves able to provide solutions to problems in open and dynamic environments. Multi-agent systems is the scientific and professional community’s current proposal to deal with distributed problems with complex solutions.

During the project work will be done to define a meta model and a meta-ontology of the complete problem, and will develop agent-based methodologies and algorithms by defining a meta-organisation from a multi-agent system that allows a negotiated solution to the problem to be found. Decision-making help tools will be proposed based on de facto standards in the world of applications development with agents, whose objective is that they are affordable for companies, and are open to allow interoperability and easy maintenance.