Modular Plant Architecture

Applied Research

MPA

Modern production industry in Europe facing the challenges of an open, dynamic and distributed market is forced to react appropriately in order to promote competitive and sustainable growth. By creating a simulation & control and immersion (applying advanced agent technology and Virtual Reality) software, MPA is providing innovative tools for decisional support.

The developed modelling concept is aimed at representing production systems of different sectors of industry as well as different hierarchy levels (e.g. plant level, workstation level) following a holarchy approach.

The simulation & control software tools are implementing PROSA autonomous agent’s architecture.

1. Production Resources

Production buildings, production segments, machinery and equipment
Flexible in order to allow agile reaction on changing boundary conditions and customer needs
Adaptive in order to render possible (re-) combination of resource modules and to increase re-usability of resources

2. Organization

Shop floor organisation
Organisational structure tailored to modular production resources
Independent organisational modules which allow (re-) combination

3. Planning

Factory planning process
Reference factory planning process to facilitate and speed up planning
Support of processes by new IT- tools (e.g. virtual reality)

Advanced agent-based manufacturing control and emulation system linked to the VR-environment via suitable interfaces and thus enables to consider dynamic aspects within the system design or plant engineering process. The manufacturing control is based on the most recently developed technology in multi-agent manufacturing control systems and holonic manufacturing systems, thus reflecting the best and latest technology available.

The control part is organized around an autonomous agent’s architecture, called PROSA (Product-Resource Order-Staff Architecture). In this paper, three types of basic agents are envisaged: order agent, product agent and resource agent.

The order agent is responsible for performing work on time and with the right quantity. The product agent holds the processing information and product knowledge (product recipe) to ensure that the correct processing on the product is performed. The resource agent is in charge of modelling the behavior of the factory resource object.

Each resource agent corresponds to a physical part and contains an information processing part that controls the resource. It holds the methods to allocate the production resource and the knowledge and procedures to organize, use and control the production resource to drive production. The resource agent is the only agent that interacts with the emulation software. Each emulation resource or entity will have a resource agent associated with it, implementing all control related tasks.

Agents control

The methodology adopted to develop the controller software has followed closely the general and wide-accepted guidelines for building a multi-agent system, including identification of agents and roles played by them, inter-agent communications and internal behavior of the agents. In contrast to the functional decomposition of classical software engineering technique, the agents are assigned to the particular entities involved in the system (e.g. resources, orders, products). This concern preserves the system’s flexibility to provide an emergent functionality as consequence of the interaction among agents.

Modern production industry in Europe faces the challenges of an open, dynamic and distributed market, so it is forced to react appropriately in order to promote competitive and sustainable growth. By creating a simulation & control and immersion (applying advanced agent technology and Virtual Reality) software, Modular Plant Architecture (MPA) is providing innovative tools for decisional support.