Adaptive Multi-Agent System for a Washing Machine Production Line

Abstract

Abstract. This paper describes the implementation of a multi-agent system in a real industrial washing machine production line aiming to integrate process and quality control, allowing the establishment of feedback control loops to support adaptation facing condition changes. For this purpose, the agent-based solution was implemented using the JADE framework, being the shared knowledge structured using a proper ontology, edited and validated in Protégé and posteriorly integrated in the multi-agent system. The solution was intensively tested using historical real production data and it is now being installed in the real production line. The preliminary results confirm the initial expectations in terms of improvement of process performance and product quality.

Agents enabling cyber-physical production systems

“…The installed solution contributes for the maximization of the factory profitability by applying self-adaptation procedures at local and global levels to face unexpected condition changes. Having this in mind, an ecosystem of autonomous agents representing the manufacturing components disposed along a production line was considered, grouped in the following types of agents according to their similarities (see [55] for more details): Product Type Agent (PTA), Product Agents (PA), Machine Agents (MA) and Independent Meta Agents (IMA). PTAs represent the catalogue of products that can be produced by the production line, i.e.…”

Intelligent products: The grace experience

“…A crucial assumption considered in this work is to keep the existing lowlevel control based on a network of Programmable Logic Controllers (PLCs) running state-of-the-art IEC 61131-3 control programs. These PLC programs ensure the real-time responsiveness and the use of the multi-agent system, located at higher control level, provides an infrastructure for the integration of process and quality control by means of intelligence and adaptation functionalities (Rodrigues et al, 2013b).…”

“…In particular, several types of agents were considered, namely the Product Type Agent (PTA), Product Agent (PA), Resource Agent (RA) and Independent Meta Agent (IMA) (Rodrigues et al, 2013b). PTAs represent the catalogue of products that can be produced by the production line and contain the process and product knowledge required to produce the product.…”

Standardization in cyber-physical systems: The ARUM case

“…For example, the GRACE project [20] developed and installed a MAS monitoring and control computational system in a factory plant producing laundry washing machines, respecting the standards. Namely, the agent-based system was developed using JADE (Java Agent Development Framework) [21], which is FIPA compliant, the exchange of messages among the agents used FIPA-ACL, and the designed ontology was expressed in the OWL language.…”