Smart Agents for the Industry 4.0 (eBook)

Foreword from Academia 5
Acknowledgement 7
Contents 9
List of Figures 15
List of Tables 19
Abbreviations 20
Glossary 27
Abstract 29
Zusammenfassung 31
1 Introduction 33
1.1 Motivation 33
1.1.1 Changing Reality – A Global Perspective 34
1.1.2 Trends and Visions of Modern Production 35
1.1.3 Digitalization and the Introduction of Cyber-Physical Systems 36
1.1.4 Challenges of the Digitalization and the Industrial (R)evolution 38
1.2 Research Goals 40
1.3 Structure of the Work 44
2 Problem Description and Fundamental Concepts 47
2.1 Strategical decision-making in a factory of the future 47
2.2 Fundamental Concepts of Industrial Manufacturing 54
2.3 Emerged Manufacturing Systems in Industrial Production 56
2.3.1 Static Control Systems in Current Factories 56
2.3.2 The Automation Pyramid and Hierarchical Production Organization 57
2.3.3 Tight Coupling in terms of Conventional Automation Infrastructures 62
2.4 The Duality of Loosely and Tightly-Coupled Systems 63
3 State of the Art 65
3.1 Global Strategies towards Future Manufacturing 67
3.1.1 Reference Architectures for an Industry 4.0 67
3.1.2 Important Research Programs targeting the Global Challenges 67
3.2 Current Approaches for Production Optimization 69
3.2.1 Organizational Optimization of the Production Process 69
3.2.2 Long-term Optimization of Production Goals using Low-Level Data 72
3.2.3 Autonomous Optimization Strategies in terms of Self-Optimization 73
3.3 Interoperability Approaches for Manufacturing Systems 74
3.4 Technical Solutions for Automated Production Control 77
3.4.1 Fieldbus Protocols 78
3.4.2 Industrial Ethernet 79
3.4.3 Publish-Subscribe within Industrial Applications 82
3.4.4 TSN and Real-Time Ethernet in Industrial Applications 85
3.5 Web Services and Service-Oriented Architectures 87
3.6 Intelligent Production Automation by Means of Multi-Agent Systems 89
3.6.1 Fundamental Concepts of Multi-Agent Systems 89
3.6.2 Communication Protocols and Standards for Multi-Agent Systems 96
3.6.3 Agent-based Approaches in Production 99
3.6.4 Agent-Based Solutions for Intelligent Manufacturing 101
3.6.5 Holonic Manufacturing Systems 103
3.6.6 Comparison of Traditional and HMS Inspired Control Solutions 105
3.7 Machine Learning in Distributed Environments 106
3.7.1 Basic Characteristics of Machine Learning 107
3.7.2 Applications of Machine Learning in Manufacturing 108
3.8 Information Modeling for Intelligent Automation 110
3.9 Advanced Interoperability Standards for Manufacturing 112
3.9.1 OPC – The Interface Standard for Generic Shop Floor Interoperability 115
3.9.2 OPC UA – The Interface Standard for Integrated Communication and Information Modeling 119
3.9.3 OPC UA System Architecture and Basic Services 124
3.9.4 AddressSpace Model and OPC UA based Information Modeling 125
3.9.5 Security Aspects of OPC UA for Reliable Information Exchange 129
3.9.6 Service-Oriented and Event-Driven Approaches through OPC UA 130
3.9.7 Interoperability with Tightly Coupled Systems by Means of OPC UA and Time-Sensitive Networking 132
3.9.8 Companion Standards and Domain-Specific Models for OPC UA 136
3.9.9 Realization of Agent-based and Decentralized Intelligence Approaches through OPC UA 139
4 Architecture of a Framework For Real-Time Interoperable Factories 145
4.1 Requirement Analysis for Legacy Systems in Automated Production Sites 146
4.1.1 Management Perspective of Automation Systems 146
4.1.2 Engineering Perspective for the Requirements of Flexible Automation Systems 148
4.1.3 Requirements from an Information Technological Perspective 150
4.2 MAS Architecture Enabling Interoperability with Existing Automation Solutions 152
4.2.1 Basic Architecture 153
4.2.2 Internal Agent Architecture 156
4.2.3 HMS Reference Architecture 160
4.2.4 Cross-Domain Architecture 163
4.3 Overall MAS Architecture for Cooperating Agents in a Smart Factory Environment 166
4.3.1 Multi-Agent System Architecture 167
4.3.2 Internal Agent Behavior and Hardware Component Interaction 171
4.4 Smart Automation of a Flexible Production by means of Evolving Software Agents 172
4.5 Limitations of the Architecture – Communication and Evolutionary Capabilities 174
4.5.1 Critical Discussion in terms of Communication Flexibility 175
4.5.2 Scalability Limitations – Learning and Evolutionary Development 176
5 Agent OPC UA – Semantic Scalability and Interoperability Architecture for Multi-Agent Systems through OPC UA 177
5.1 Bridging the Gap to OPC Classic 180
5.2 Information Modeling and Infrastructure for CPPS 181
5.3 Implementation of OPC UA Enabled Multi-Agent Systems 183
5.3.1 Agent Requirements Specification 184
5.3.2 Multi-Agent System Requirements Specification 185
5.3.3 Required Functionalities of an OPC UA enabled MAS 186
5.3.4 Messaging System 186
5.3.5 Data Storage 196
5.3.6 Reading Values from Remote Devices 200
5.3.7 White Page Services 205
5.3.8 Yellow Page Service Realization 207
5.3.9 Discovery Process in MAS and Dynamic Networking 208
5.3.10 Incorporation of Traditional MAS through Gateway Agents 209
5.3.11 Interconnection of Multiple MAS by means of Mediation Agents 212
5.4 Representation of Intelligent Agents by means of OPC UA 214
5.4.1 OPC UA AddressSpace Representation for Smart Agents 215
5.4.2 OPC UA AgentType Model 217
5.5 Semantic Integration of Agent Communication and Interoperability 225
5.5.1 The MessageType Object Definition 225
5.5.2 Hierarchical Organization of Message Objects 228
5.5.3 Compatibility of OPC UA to Legacy ACL Messages 229
5.5.4 The ContentType Specification 230
5.5.5 The AbilityType Specification 231
5.5.6 SensorType Definitions 232
5.5.7 Overall Information Model for OPC UA Based MAS 232
6 Management System Integration of OPC UA based MAS 235
6.1 Architecture Extension for Decentralized Planning 235
6.1.1 General Capabilities/Requirements for Resource Planning 236
6.1.2 The ERP Agent as Gateway to High-Level Planning Systems 237
6.1.3 Planning and Scheduling Capabilities 240
6.1.4 Decentralized Organization Based on a Blackboard Approach 241
6.2 Incorporation of an ERP System Into OPC UA Based MAS 242
6.2.1 The ERP Information Model Specification 243
6.2.2 Incorporation of an Open Source ERP System 245
6.2.3 Cooperative Reactive Production Planning 246
7 Flexible Manufacturing based on Autonomous, Decentralized Systems 247
7.1 Learning Agents for Flexible Manufacturing 247
7.2 Machine Learning in Learning Agent Environments 249
7.3 Predictive Maintenance Manufacturing Scenarios 250
7.3.1 Prediction of the Remaining Useful Lifetime 251
7.3.2 Direct RUL Prediction with Neural Network Approaches 252
7.4 Application of ML Scenarios in MAS 253
7.4.1 Intelligent Negotiation Techniques in MAS 253
7.4.2 Quantitative Price Function for Negotiating Agents 255
7.4.3 Predictive Maintenance Costs through Machine Learning 257
7.4.4 Implementation of the RUL Prediction into Smart Agents 261
7.5 RUL Learning Agents in OPC UA Based MAS 262
8 Use-cases for Industrial Automation Processes 263
8.1 Domain Ontology for the “myJoghurt” Testbed 264
8.1.1 myJoghurt Demonstration Scenario and Work Flow 264
8.1.2 OPC UA Information Model for the Mapping of Agent Communication 268
8.1.3 Achievements and Limitations of the Use-Case Design Model 275
8.2 Manufacturing Use-Case 275
8.2.1 Process Description 276
8.2.2 Autonomous Organization of the Production Process 276
8.3 Demonstration Scenario – The Industry 4.0 Testbed 282
8.3.1 Technical Setup and Realization of the Demonstrator 282
8.3.2 Simulation of the Manufacturing Process 285
8.3.3 Learning and Evolution of the Software Agents 288
9 Future Research Topics 289
9.1 Semantic Scalability for Domain Specific Use-Cases 289
9.2 Generic Extensibility of Communication Concepts Based on Ontologies 290
Summary 292
Bibliography 294
Appendices 325
A Protocols for Production Automation 326
A.1 History and Background on Fieldbus Systems in Modern Manufacturing 326
A.2 Basics and Historical Background of Industrial Ethernet 332
B Architecture and Technical Realization of the MAS 337
B.1 Multi-Agent System Architecture 337
B.2 OPC UA Based MAS 338
B.2.1 GenericContentType Messages 338
B.2.2 ContentType Subtype Mapping 339
B.2.3 AbilityType Filtering Feature 340
B.2.4 ERP Model Extension 340
B.2.5 Technical Setup of the Demonstrator 342
C Machine Learning Results 344
C.1 Initial Prediction and Results of Genetic Algorithms 344