Upload
semanticsconference
View
122
Download
0
Embed Size (px)
Citation preview
KIT – University of the State of Baden-Württemberg andNational Laboratory of the Helmholtz Association
KARLSRUHE SERVICE RESEARCH INSTITUTE (KSRI)
www.kit.edu
Semantic Technologies for Assisted Decision-Making in Industrial Maintenance
Sebastian BaderResearch Associate
Institut KSRI05/03/20232 Sebastian Bader
Predictive Maintenance• Forecasting break-down probabilities
Condition-Based Maintenance• Discover failure patterns
Preventive Maintenance• Specified service intervals
Reactive Maintenance• Run to failure
Industrial Maintenance Process
!
Amount of unplanned downtimes
Institut KSRI05/03/20233
Improvement Areas
Sebastian Bader
Dispatcher
Client
TechnicianMachine
Remote support
Schedule
Tour
Local/global planning
Real-time tour optimization
Predictive Maintenance
Information provision
Semi-automateddecision making
Institut KSRI05/03/20234
Next Generation of Maintenance
Reduction of unplanned downtimes
Less travel time for field technicians by tour optimization
Improved planning of resources and capacities
Automated/Supported decision making where possible
Automatic data exchange with customers/suppliers
Integrating external services and competences Provisioning of contextualized information
Sebastian Bader
Institut KSRI05/03/20235
Challenges
How can advanced data insights be used to create business value?
How can available data contribute to a more efficient maintenance process?
What are the current limitations and how can we solve them?
Sebastian Bader
Institut KSRI05/03/20236
Predictive Analytics provides flexibility… … to prepare resources … to organize technicians … to adjust capacities and demands
Data-driven approaches reduce complexity… … by regarding all side effects … by suggesting appropriate actions … by supplying related information
Transforming Predictive Analytics into Business Value
Sebastian Bader
DispatcherSchedulePredictive
Maintenance
Capacity
Demand
Predictions at its own are not sufficient, only the ability to react provides value!
Reducing uncertainty increases efficiency:
Therefore, an integrated support system for the whole process is necessary.
Institut KSRI05/03/20237
System Integration via Semantic Web Technologies
Current systems already solve some challenges forecasting machine downtimes optimized scheduling of technicians real-time tour planning
Need for addressing constantly added/removed resources New machine instances, types, technologies New customers, departments, partners Disconnected machines, expiring contracts
Need for system integration across departments, organizations, and countries
Need for flexible, modularized and decentralized integration approach
Sebastian Bader
TourSchedulePredictive
Maintenance
Institut KSRI05/03/20239
System Integration via Semantic Web Technologies
How to enable the integration of external services with potentially unknown requirements, heterogeneousaccess methods and varying data formats into a decentralized network?
Smart Web Services1 (SmartWS) Encapsulate context-based decision logic Lifting and lowering to agreed data format according to Linked Data Principles Access via HTTP and REST Self-describing and therefore automatically
controllable Consumer and producer at the same time
(=Prosumers)
Sebastian Bader
TourSchedulePredictive
Maintenance
System 1
System2
HTTP REST
RDF
Wrapperlibrary
Wrapperlibrary
Lifti
ng
LoweringJSON
mapping mapping
Output Functionality InputProvenance
2 Maleshkova, Maria, et al. "Smart Web Services (SmartWS)–The Future of Services on the Web." The IPSI BgD Transactions on Advanced Research: 15.
Institut KSRI05/03/202310 Sebastian Bader
Reusable SmartWS
Data Sources, Devices, Sensors, Wearables, Algorithms, etc.
Composite Applications
SmartWS
Devices
SmartWS
Sensors
SmartWS
Algorithms
SmartWSSmartWS SmartWS
Execution Engine
Reference SmartWS Architecture
Institut KSRI05/03/202311
Web Services and Linked Data Platform
Access to data Stored, managed and published through DBs
Linked Data Platform2 for reading/writing RDF RESTful methods for data requesting and manipulation
SmartWS provide Linked APIs with semantic descriptions
Requesting Web services WSDL/SOAP or RESTful communication
Sebastian Bader
Consistent handling of data and services
2 Speicher, Steve, John Arwe, and Ashok Malhotra. "Linked data platform 1.0." W3C Recommendation, February 26 (2015).
Institut KSRI05/03/202312
Provision of Contextualized Information
Identify topics and context Reports, manuals, posts
Understand the current situation Dynamic information from heterogeneous
input channels Static knowledge on processes and
resources
Modeling information objects as resources, enhanced with meta data, in a common manner
Sebastian Bader
Technician
Machine
History
Task
Situation
Institut KSRI05/03/202313
Social Maintenance Network
“There must be someone who knows the solution to my problem. How can I find him? How can I access his expertise?” Implicit knowledge not queryable Segregation by organizational unit, language, region, …
1. Connect people depending on qualification, experience, task, and availability
2. Supply available information where needed
Solution:Social network for fast and reliable communication and adaptive information provision
Sebastian Bader
Dispatcher Technician
Institut KSRI05/03/202314 Sebastian Bader
Platform for information and knowledge exchange based on Linked Data representations
Semantic Media Wiki
Semantic MediaWiki
• Collaborative work• Sharing knowledge• Easy syntax• Browser-based (stationary and mobile)• Perfect integration with semantic
technologies• Access on data views (near real-time)• OLAP functionality• Extendable platform
Institut KSRI05/03/202315
Semantic Text Analysis and Similarity Matching
From Semantic Media Wiki to Social Platform
Sebastian Bader
TaskRouteChatHelpTools
Mobile application
Task X Machine Y
Task-related information views
Activity 1
Activity 2
Task A
Problem P
ID: 0053A435-ZD
Changing air filter of AC unit
Type:CutterInstalled: 2011Color:greenLocation:Tech Inc.Configuration:DFR-24
Mario RossiJohn Doe
Max MustermannJean Untel
Community support
Chat functionality
Procedure:1. Open shell2. Check power supply3. Change fuse4. Start test sequence5. Check power LED6. Detach wires7. Lift filter8. Insert new filter9. Attach wires10.Restart test sequence11.Fill report12.Let customer sign13.Close shell14.Start machine
History:Oil pressure errorVibrationsRegular maintenanceInstallation
Client:Name:
Tech Inc.Contact: Peter MüllerTel. no.: 01234 555Time: 9:00 to 11:30Address: IoT Road 1
Smallville
Institut KSRI05/03/202316 Sebastian Bader
MAINTENANCE SCENARIO BUSINESS MODELS
CUSTOMER (LEASING)
Leasing inclusive repair commitment
MANUFACTURER
F CUSTOMER (MACHINE OWNER)
Full-Service-Contract
MANUFACTURER/MAINTAINER
PLATFORM
@SENSOR DATA
(periodic intervals)
BREA
KDO
WN
PRE
DEC
TIO
NBREAKD
OW
N PRED
ECTION
;ANALYTIC RESULTS
component breakdown probability etc.
SENSOR DATAmeasurements, conditions etc.
2PREDICTIVE ANALYTICS
measurements, conditions etc.
IMPROVEMENTS
INCREASING EFFICIENCYShorter maintenance and travel times
INCREASING AVAILABILITYMinimizing unexpected breakdowns
MINIMIZING MAINTENANCE COSTSReduced investigation time
MAXIMIZING TOTAL LIFETIMEOptimized maintenance
3@
Institut KSRI05/03/202317
Future Business Cases
Full-Service Contracts Automated maintenance organization allows efficient risk
management Machine-as-a-Service instead of single sales event
Strategic skill management Integrated modules enable the detection of missing/required skills
of work force Combination of operational planning with strategic simulations lead
to fact-based decisions
Externalization of low profit tasks Marketplace for external maintenance provider Gradual access to sensitive technical information
Sebastian Bader
Institut KSRI05/03/202318
Conclusion
Semantic Web Technologies enable a flexible and decentralized integration of heterogeneous resources.
Consistent data modeling with RDF for a system-wide information access
Smart Web Services encapsulate automated decision logic in order to reduce complexity and increase processing speed
Semantic annotations of documents, situations, and employees allow context-related information provision
Semantic Technologies enable more efficient industrial maintenance processes with new business models
Sebastian Bader
Institut KSRI05/03/202319
Acknowledgements
This work is partially supported by the German Federal Ministry for Economic Affairs and Energy (BMWi) as part of the “Smart Service Welt” program under grant number 01 MD16015 B (STEP)
Sebastian Bader