Design and Implementation of a Smart, Product-Led Production Control Using Industrial Agents

In theory, the design of modern production systems in the form of a cyber–physical production system (CPPS) allows more flexibility, simple expandability, quick adaptability, and intelligent production control by the product. Multiagent systems (MASs) are thereby recommended as control solution because of their autonomy and dynamic decentralized architecture. Although their potential use and technical excellence have been proven, the costs of implementation and maintenance still outweigh their supposed advantages. This results in low acceptance and usage of operational MAS in the industry. This article describes topics that need to be considered when designing and implementing an MAS as production control for a CPPS in the context of customized mass production. Generally developed approaches are presented, which were implemented in a commercially developed agent framework and validated on the basis of an industrial use case for a product-led filling process in lot size one. All implemented concepts aim to be reusable in comparable applications across industries. In combination with the MAS-internal testing approach also presented, this should contribute to faster, more cost-effective implementation of reliable MAS solutions and ultimately increase their technical maturity.