Smart Agents for the Industry 4.0

Enabling Machine Learning in Industrial Production

verfasst von: Max Hoffmann

Verlag: Springer Fachmedien Wiesbaden

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten

Über dieses Buch

Max Hoffmann describes the realization of a framework that enables autonomous decision-making in industrial manufacturing processes by means of multi-agent systems and the OPC UA meta-modeling standard. The integration of communication patterns and SOA with grown manufacturing systems enables an upgrade of legacy environments in terms of Industry 4.0 related technologies. The added value of the derived solutions are validated through an industrial use case and verified by the development of a demonstrator that includes elements of self-optimization through Machine Learning and communication with high-level planning systems such as ERP.

About the Author:

Dr.-Ing. Max Hoffmann is a scientific researcher at the Institute of Information Management in Mechanical Engineering, RWTH Aachen University, Germany, and leads the group “Industrial Big Data”. His research emphasizes on production optimization by means of data integration through interoperability and communication standards for industrial manufacturing and integrated analysis by using Machine Learning and stream-based information processing.

Inhaltsverzeichnis

Frontmatter

Chapter 1. Introduction

Abstract

Recent technological advancements from customer sectors become an integral part of societal and economical reality. Data analytics, the rise of artificial intelligence and big data are major trends in all sectors of modern life. Especially industrial production is significantly affected by the changes that are implied through an application of these forecasting technologies.

Max Hoffmann

Chapter 2. Problem Description and Fundamental Concepts

Abstract

This chapter deals with the fundamental concepts of current manufacturing systems and problems need to be overcome on the way to a factory of the future. The first step to approach these challenges consists in a characterization of paradigm shifts that are needed to address the information transparency demands of a factory that is able to reconfigure itself by means of up-to-date information from the field. Fundamental concepts of modern industrial manufacturing deliver a brief description of industrial manufacturing prior to a basic characterization of emerged manufacturing systems in current factories.

Max Hoffmann

Chapter 3. State of the Art

Abstract

The first part of the state of the art section is focused on general trends and requirements on the way to a factory of the future. Current strategies and paradigms to optimize automated production processes are pointed out first, before further parts of this section will go into detail regarding technical realization by means of concrete technology stacks.

Max Hoffmann

Chapter 4. Architecture of a Framework For Real-Time Interoperable Factories

Abstract

Considering the challenges that are connected to the complexity of current manufacturing systems as pointed out in the problem description and fundamentals, it is the aim of this chapter to outline an architectural approach that takes into account these requirements. The challenges in terms of the derivation of an agent-based/HMS inspired architecture regarding novel manufacturing systems are very well pointed out by Shen and Norrie (1999), Jin-Hai et al. (2003), McFarlane and Bussmann (2003) and Botti and Giret (2008): enterprise integration, distributed organization, heterogeneous environments, interoperability, open and dynamic structures, cooperation, integration of humans with software and hardware, agility, scalability, and fault tolerance. The boundary conditions made up by these challenges as well as the technological approaches pointed out in the state of the art section of this thesis constitute the cornerstone for this architecture.

Max Hoffmann

Chapter 5. Agent OPC UA – Semantic Scalability and Interoperability Architecture for Multi-Agent Systems through OPC UA

Abstract

The architectural approaches derived in the previous chapter pointed out sine key concepts to enable flexible decision making by means of agentbased, self-organizing and decentralized systems. Means to realize such architecture in practical applications were pointed out despite the limitations that have to faced with regard to current state-of-the-art solutions for the communication in such dynamic networks. Furthermore, a critical discussion of the conventional information exchange methodologies o ered some fundamental weaknesses in terms of semantic scalability, extensibility and flexibility in terms communication within MAS.

Max Hoffmann

Chapter 6. Management System Integration of OPC UA based MAS

Abstract

The incorporation and mapping of multi-agent systems by means of the OPC UA standard opens up various potentials with regard to a flexible production automation. One major benefit of the extended ICT lies in the interconnection of intelligent software agents with high-level planning systems in terms of vertical integration. These way, the advantages of applying both, centralized planning approaches within top-level systems and at the same time decentralized reconfiguration, can be exploited in the automated production environment.

Max Hoffmann

Chapter 7. Flexible Manufacturing based on Autonomous, Decentralized Systems

Abstract

In multi-agent systems applications, different types of agents can be distinguished, such as simple reflex agents, utility based agents and learning agents (Russell et al., 2010). Current applications of MAS are often characterized by simple reflex agents that exhibit a library of predefined rule sets. These agents have a determined behavior and are comparatively easy to set up, however they are also characterized by limited intelligence.

Max Hoffmann

Chapter 8. Use-cases for Industrial Automation Processes

Abstract

The use-case section of this work demonstrates the potentials of the derived framework in terms of real-world applications that make use of the scalability concepts of OPC UA based MAS. One major aim of these applications is to show the extensibility of the derived framework in terms of domainspecific information models. The applicability to solve practical problems that are of interest in production and manufacturing organization is shown in terms of two use-cases, of whom one resulted in the development of an Industry 4.0 testbed that was demonstrated at the Hannover Fair 2016.

Max Hoffmann

Chapter 9. Future Research Topics

Abstract

This work has demonstrated an information modeling approach that is able to describe, integrate and optimize data and information flows within current automation systems of modern factories. As emphasized in the usecase section of this thesis, a decisive factor to strive for benefits of such approach consists in the incorporation of domain-specific knowledge into the development of automated process.

Max Hoffmann

Backmatter

Titel: Smart Agents for the Industry 4.0
verfasst von: Max Hoffmann
Verlag: Springer Fachmedien Wiesbaden
Electronic ISBN: 978-3-658-27742-0
Print ISBN: 978-3-658-27741-3
DOI: https://doi.org/10.1007/978-3-658-27742-0