A Modular and Low-Cost Infrastructure for Industrie 4.0 Applications Linking Real-Time Data to the Cloud T. Ortmaier, I. Maurer, M. Riva, C. Hansen Institute of Mechatronic Systems (imes) Leibniz Universität Hannover Appelstraße 11 A 30167 Hannover E-Mail: Tobias.Ortmaier@imes.uni-hannover.de Internet: www.imes.uni-hannover.de Telefon: +49 511 762 4179
A Modular and Low-Cost Infrastructure for Industrie 4.0 Applications Structure Introduction Concept Infrastructure Architecture Model Factory Exemplary Use-Cases Summary Seite 2
Institute of Mechatronic Systems (imes) in the Context of Industry 4.0 Industry 4.0 creates what has been called the "smart factory, containing: data integration of all involved components (variety), real-time data acquisition (velocity), analysis of large volume data sets (volume). Big Data [1] and enabling several goals and methods: real-time condition monitoring, predictive maintenance, energy management, process optimization, combining identification and parameterization methods, and many more. Apps This requires an infrastructure to collect, store, and process (big) data in real-time: under development in the new research group Integrated Systems & Machine Learning, tested on the fully automated handling process in a laboratory model factory. Infrastructure Seite 3
Concept Automation Pyramid Processed / evaluated data Main data flow (Process data) Private cloud Enterprise level Production planning ERP Company Server Plant management level Production control, operational data management MES Operation data management (Process-) control level Human Ressource Interface SCADA Operating- and Monitoringsystems Operation level Machine and plant control SPS Programmable Logic Controller (PLC) Microcontroller Field level In- / output signals Sensors / actuators Servo drives, Field devices Process level Manufacturing / production process Automated production plant Own representation based on [2] Seite 4
Concept Design Goals and Apps Transparency Data security Low-cost solution Modularity / Scalability Monitoring Availability increase Predictive maintenance Seite 5
Infrastructure Architecture Conceptual Setting Wide range of possible analysis methods Robotics Energy, System Identification, Structural Health, Plant Performance, Economic Efficiency, etc. Plant / Process data Set points, Sensor values etc. Embedded System Data Acquisition, Preprocessing, Filtering, Anonymisation etc. Data Analytics Platform Storage, Management, Data Analysis, Modeling, Optimization etc. Interactive Web Services Visualization (GUI) Parameterization Trigger alerts, optimize processes Save results, parameterization Seite 6
Embedded System Infrastructure Architecture Server Architecture High performance [3] Dashboard OPC UA real-time publish subscribe messaging system [4] HTTP(S)-Streaming Real Time Analytics Storm HBase NoSQL Can handle terabytes of data without performance impact [5] HTTP(S) HTTP(S) Development Low latency Hive SQL Controlling HTTP(S) / OPC UA HDFS Storage Save and backup of large volume of process data [3] HTTP(S) Seite 7
Model Factory Seite 8
Model factory Structure Programming interface Higher-level control / logic system: p500 SCARA Delta Robot & Belts 6-Axis-Robot + linear axis Stacker crane - PLC 3200C - Brake resistor - 4 servo axes - PLC 3200C - Brake resistor - 4 + 2 servo axes - PLC 3200C - Brake resistor - 6 + 1 servo axes - PLC 3200C - Brake resistor - 3 servo axes Seite 9
Exemplary Use-Cases Two Exemplary Use-Cases Energy Monitoring / Process optimization Application: model factory Delta robot (4 axis) Stacker crane (3 axis) 2 conveyor belts Goals: Quantification of energy consumption (on module / component basis) Identification of potentials to increase efficiency and optimize processes Challenge: Synchronize and analyze data of different sources / components Processing and storing large amount of realtime (1kHz) process data Condition Monitoring / Predictive Maintenance Application: stacker crane 3 axis Goals: Diagnosis / system condition monitoring Anomaly detection and classification (e.g. based on bearing damage or friction) Challenge : Modelling based on process data (data science methods) Online anomaly detection and classification of real-time (1kHz) process data-streams Seite 10
Exemplary Use-Cases Energy-Monitoring / Process Optimization Analysis: Energy-Monitoring / -Management [6,7,8] Quantification of energy consumption of plant and individual components Detection of excessive energy losses and peak loads Development of generic approaches to use for any type of robot Goals: Energy minimization through process optimization Identification of potentials to increase efficiency Assessment of energy demands for alternative drive components or energy supply concepts (e.g. intermediate circuit) Optimization: Energy optimal motion planning / task synchronisation Recommendation for action (DC networking, recovery, energy storage, component replacement, ) power Stacker crane Delta robot Belt 1 Belt 2 time time Resolution: up to 1kHz Seite 11
Exemplary Use-Cases Condition Monitoring / Predictive Maintenance Condition Monitoring: Real-time analysis and diagnosis Monitoring as a decision-making basis for component replacement / error handling Methods: Principal component analysis (PCA) k-nearest-neighbor (knn) Trainings-data sets without errors with errors (only for classification) Detection of anomalies versus training data Classification of errors (e. g. belt slippage) Controller needs >2s for belt slippage detection New: belt slippage detection in 0,5s belt slippage Optimization: Switching of control trategies (e.g. emergency stop) Predictive component replacement before damage occurs Example 1: no slippage Example 2: slippage (fast) Example 3: slippage (slow) Seite 12
Industrie 4.0 Applications Linking Real-Time Data to the Cloud Summary imes is researching the topic of Industrie 4.0 in industrial applications, including: Development and implementation of a modern big data-infrastructure: real-time data aggregation, data preprocessing, data handling for big data-applications, and visualization / controlling methods. Research of data science / machine learning methods for process / plant data analysis, e. g. Energy monitoring, Process optimization, Condition monitoring etc. Focus on online / real-time process and plant data-stream analysis. Seite 13
Thank you for your attention. T. Ortmaier, I. Maurer, M. Riva, C. Hansen Institut of Mechatronic Systems (imes) Leibniz Universität Hannover Appelstraße 11 A 30167 Hannover, Germany mail: Tobias.Ortmaier@imes.uni-hannover.de web: www.imes.uni-hannover.de phone: +49 (0)511-762 - 4179
A Modular and Low-Cost Infrastructure for Industrie 4.0 Applications References Literature: [1] Mayer-Schönberger V. and Cukier K. (2013), "Big data: A revolution that will transform how we live, work, and think" Houghton Mifflin Harcourt. [2] VDI/VDE-Gesellschaft Mess und Automatisierungstechnik (GMA). "Cyber-Physical Systems: Chancen und Nutzen aus Sicht der Automation". Thesen und Handlungsfelder, April 2013. [3] Shvachko K., Kuang H., Radia S. and Chansler R. (2010), "The Hadoop Distributed File System", In 2010 IEEE 26th Symposium on Mass Storage Systems and Technologies (MSST)., May, 2010., pp. 1-10. [4] Kreps, J., Narkhede, N., & Rao, J. (2011). "Kafka: A distributed messaging system for log processing". In Proceedings of the NetDB (pp. 1-7). [5] Aydin G., Hallac I.R. and Karakus B. (2015), "Architecture and implementation of a scalable sensor data storage and analysis system using cloud computing and big data technologies", Journal of Sensors. Vol. 2015 Hindawi Publishing Corporation. [6] Hansen C., Öltjen J., Meike D. and Ortmaier T. (2012), "Enhanced Approach for Energy-Efficient Trajectory Generation of Industrial Robots", Proceedings of the 2012 IEEE International Conference on Automation Science and Engineering. [7] Hansen C., Kotlarski J. and Ortmaier T. (2014), "Optimal motion planning for energy efficient multiaxis applications", International Journal of Mechatronics and Automation. [8] Hansen C., Eggers K., Kotlarski J. and Ortmaier T. (2015), "Concurrent Energy Efficiency Optimization of Multi-Axis Positioning Tasks", The 10th IEEE Conference on Industrial Electronics and Applications (ICIEA 2015). Figures: KUKA AG, Lenze SE, Dell Technologies Inc., Apache Software Foundation, GINO AG Seite 15