Overview of an Embedded Platform for Big Data in Manufacturing
This presentation was given at JavaOne Embedded 2012 in San Francisco. The talk goes over what is needed in an embedded platform to enable big data analytics in manufacturing.
Manufacturing Big Data Analytics Monitor à Detect à Predict Realtime Analysis of Integrated Manufacturing Systems Machine Learning for Historical Analysis Analytics Platform system insights Thursday, October 4, 12
Manufacturing is Big: $2 Trillion sector Discrete Manufacturing: Products for consumers and the supply chain High potential for productivity improvement: Metal cutting equipment spend less than 25% usefully Manufacturing generates a very large amount of data: Most of this falls on the floor Thursday, October 4, 12
transformation How well are we making this transformation? Productivity? Profitability? Return on asset? Part quality? Employee safety? Product reliability? Sustainability? Energy usage? Pollution? Thursday, October 4, 12
Large facilities with disparate types of equipment Limited automation: Human driven and controlled Very low computational power. Average equipment > 10 years old Unintegrated “islands of excellence” http://www.michsci.com/Services/manufacturing.htm Thursday, October 4, 12
scales can vary from µ-seconds to days anytime: process management neartime: process improvement realtime: process control m-Seconds Seconds Hours Days Process Interface Sub- Components Manufacturing Equipment Manufacturing Supply Chain Manufacturing Enterprise Temporal Decision Scale Manufacturing Analysis Scale Thursday, October 4, 12
vimana Transfer Station Transfer Station Part Part Part Process Station Process Station Realtime Part Monitoring and Process History OEM Sourcing Metrology Station Supply Chain Management Long Term Regulatory Records Thursday, October 4, 12
Patterns Feedback and Refinement Process Planning Process Execution Process-based Reasoning Continuous Improvement with vimana Energy Monitoring using Java SE Embedded Process Advisory Services Energy Efficient Execution Energy Management Thursday, October 4, 12
Management Framework Data Collection Data Storage Data Analysis Big Data Apps Efficiently capture and deliver data from the manufacturing equipment to the collection systems Perform reliable realtime analysis using statistical machine learning (SML) and complex event processing (CEP) Robustly collect and store data in high- speed databases Thursday, October 4, 12
Platform for manufacturing productivity management and improvement High Speed Data Bus High Speed – High Availability Data Capture User Facility User Facility User Facility CRE: Pattern Matching and Anomaly Detection High Speed Process Execution Database MLE: Learning and Refinement 3rd Party Apps User Data Anonymizer User Interfaces vimana Platform Thursday, October 4, 12
Insights, 2012 Productivity Monitoring vimana user-interface Keeps track of time spent by the device in production activities and duration of production losses Auto-classify production losses into reasons, including: setup, breakdowns, tool change, program adjustment. Data Sources for Auto- Classification: •CNC, PLCs, Sensors •User Input, including Device and Plant Schedule Supported Metrics: •Production Efficiency •Part Count •Cycle Time •OEE •Energy Usage Thursday, October 4, 12
of standardized data sources in the shopfloor: complex, disparate types of equipment with proprietary interfaces Biggest gap in enabling end-to-end connectivity and analysis standards? Thursday, October 4, 12
standard •Based on open protocols •Extensible and lightweight What is MTConnect? • Incompatibilities in protocols • Incompatibilities in data formats • Hard to configure/set up • Proprietary silos of technology Addressing Problems Open standard defining protocol and interchange format to facilitate communication between devices in manufacturing systems •Uses standardized tags in schema •Lightweight: can act as enabler for other standards •Unintrusive: Does not assume any business logic •Simple: Enables connectivity in legacy equipment MTConnect and Other Standards Thursday, October 4, 12
SE Embedded powered devices for data collection across the shopfloor Device-agnostic integration with CNCs, PLCs, Sensors, Power Meters, and other data sources Leverage MTConnect open standard for data interoperability Realtime and historical analysis using vimana cloud-based platform for big data Manufacturing Shop Floor System Insights vimana Cloud Device 2 Device 1 Java SE Embedded Technology Device ... Device n Analysis Engine Device- Specific Adapters MTConnect Agent MTConnect Thursday, October 4, 12