Transcript of Manufacturing Analytics at Scale
- 1. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and affiliates.
All rights reserved. Manufacturing analytics at scale 1 Soundar
Srinivasan Bosch Data Mining Services and Solutions, Palo Alto, CA
Jeff Thompson, Ruobing Chen, Juergen Heit, Dirk Slama
- 2. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and affiliates.
All rights reserved. Bosch@Data Science Summit, 2015 2 Outline
Bosch overview Core business sectors World class manufacturing Data
mining at Bosch Successful applications in manufacturing Unique
challenges encountered Need for further research
- 3. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and affiliates.
All rights reserved. Bosch@Data Science Summit, 2015 3 2014 key
figures One of the worlds largest suppliers of automotive
technology Industrial Technology Energy and Building Technology
Bosch Group 48,9 billion euros in sales R&D investment: 4.9
billion euros 360,000 associates as per April 1.15* Mobility
Solutions Leading in drive and control technology, packaging, and
process technology Leading manufacturer of security technology
Global market leader of energie-efficent heating products and
hot-water solutions Consumer Goods Leading supplier of power tools
and accessories Leading supplier of household appliances 68% share
of sales *including BSH Hausgerte GmbH (formerly BSH Bosch und
Siemens Hausgerte GmbH) and Robert Bosch Automotive Steering GmbH
(formerly ZF Lenksysteme GmbH). 32% share of sales
- 4. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and affiliates.
All rights reserved. Bosch@Data Science Summit, 2015 4 Four
business sectors: A diverse, and rich field for data science
applications Mobility Solutions Industrial Technology Energy and
Building Technology Consumer Goods
- 5. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and affiliates.
All rights reserved. Bosch@Data Science Summit, 2015 5 Freely
programmable control units PLC and PC based control units Field bus
(ethernet- based) Flexible production systems Digital data storage
Usage of internet standards Integrated IP- connection Identifiable
and communicating objects Mobile operation Scalable systems (cloud
as storage, ..) Self-optimising systems Internet-of-things Advanced
manufacturing: The next industrial revolution Industry 1.0 2.
industrial revolution 3. industrial revolution 4. industrial
revolution Industry 2.0 Industry 3.0 Mech. control (cam disc, cam)
Energy: water / steam power Punch cards as program memory Conveyer
belts Master shafts Energy: electrical 1. industrial revolution
Mechanisation Electrification Digitalization Connectivity and
Traceability The transformation of Industry 3.0 to Industry 4.0
(advanced manufacturing) occurs gradually Industry 4.0
- 6. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and affiliates.
All rights reserved. Bosch@Data Science Summit, 2015 6 Two
perspectives of Bosch in advanced manufacturing Technology and
solution supplier for OEMs and end users LEAD PROVIDER System
manufacturer view / production resource view LEAD OPERATOR Product
manufacturer view / product view First mover in the realisation of
integrated concepts with equipment providers Big Data Business
processesDecentralised intelligence Machine models Software Added
value networks Connection Production models
- 7. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and affiliates.
All rights reserved. Bosch@Data Science Summit, 2015 7 As of
12/2014 200+ Manufacturing facilities 1000s of assembly lines
Billions Of parts manufactured each year Bosch manufacturing
- 8. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and affiliates.
All rights reserved. Bosch@Data Science Summit, 2015 8
Manufacturing use cases Test and Calibration Time Reduction
Prediction of test results Prediction of calibration parameters
Quality Improvement Descriptive analytics for root-cause analysis
Early prediction from process parameters Self-optimizing assembly
line Warranty Cost Reduction Prediction of field failures from Test
and process data Cross-value stream analysis Yield Improvement
Benchmark analysis across lines and plants Pin-point possible root
causes for performance bottlenecks (OEE, cycle time) Predictive
Maintenance Identify top failure causes Predict component failures
to avoid unscheduled machine down-times
- 9. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and affiliates.
All rights reserved. Bosch@Data Science Summit, 2015 9 Business
Objective: Reduce test and calibration time in the production of
mobile hydraulic pumps Impact Example: Test time reduction 35%
reduction in test and calibration time via accurate prediction of
calibration and test results
- 10. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and
affiliates. All rights reserved. Bosch@Data Science Summit, 2015 10
Example: Test time reduction Problem: Bottleneck Test Benches
Approach: 1) Identify candidate tests for removal 2) Identify test
groups run in parallel 3) Use feature selection methods (group
Lasso) to identify least important test measurements. 4) Remove
least important test measurements (saving test time) 5) Train a
predictive model to predict test outcome from remaining
measurements. Layout of the assembly line
- 11. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and
affiliates. All rights reserved. Bosch@Data Science Summit, 2015 11
Scalable Group Lasso min 1 =1 log 1 + 0 + =1 , + =1 We used
Limited-memory BFGS (L-BFGS) with Block Coordinate Descent (BCD) to
solve the optimization problem. L-BFGS is used to obtain a
quadratic approximation of the logistic regression. BCD is used to
solve the resulting sub-problem, i.e., a quadratic problem with
group penalty.
- 12. 12 G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and
affiliates. All rights reserved. Bosch@Data Science Summit, 2015
Scalable Group Lasso Three parts of the approach can be distributed
Gradient computation of the logistic function Storage of L-BFGS
history BCD sub-problem solver: minimize each block simultaneously
When to and why distribute? Distributing the gradient computation
is beneficial when sample size is large Distributing the storage of
L-BFGS history is beneficial when there are a lot of features Chen
et al., (NIPS 2014) show that this distributed version is
advantageous over the original only when the number of feature is
larger than 10Mil. Distributing BCD is beneficial only when the
number of groups is large
- 13. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and
affiliates. All rights reserved. Bosch@Data Science Summit, 2015 13
Analytics and production environment Device Management Device
Abstraction Event Management Software Provisioning Identity
Management Production Env.Analytics Environment Hadoop MongoDB DB
Connectors Custom Scripts SAS IBM SPSS Statistica Alpine KNIME
Revolution R RapidMiner Extraction, Trans- formation, Loading
Aggregate Data Historic Training Data Analytics, Machine Learning
Descriptive Analysis Predictive Model Extraction, Transformation
Predictive Model Prognosis, Decision (-Support) Sales Data
Production Data Warranty Data Device Data
- 14. Challenges in predicting defects in manufacturing 14 Large,
but distributed data E.g. One product variant in one plant ~15
million units, 29 data sources, 17 TB data, 22 billion measurements
High dimensional 100s-1000s typical Schema- and
dictionary-migration over time Near real-time and
resource-constrained deployment G1/PJ-DM | 7/17/2015 | 2015 Robert
Bosch LLC and affiliates. All rights reserved. Bosch@Data Science
Summit, 2015
- 15. 15 G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and
affiliates. All rights reserved. Bosch@Data Science Summit, 2015
Other data science challenges in manufacturing Data is short-term
stationary Time and feature correlation Label noise Very low (but
costly) incidence rates 0-few ppm typical Unequal costs of false
alarms and false negatives High accuracy and quality
requirements
- 16. 16 G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and
affiliates. All rights reserved. Bosch@Data Science Summit, 2015
Need for expanding research in manufacturing IEEE Big Data for
Advanced Manufacturing Workshop 2015 IEEE International Conference
on Big Data Oct 29 Nov 01 2015 @Santa Clara, USA
http://ieeebdam15.stanford.edu/
- 17. G1/PJ-DM | 7/17/2015 | 2015 Robert Bosch LLC and
affiliates. All rights reserved. Bosch@Data Science Summit, 2015
Backup 17 Advanced manufacturing App Store/Digital Services (2)
Connected Products (1) 3D Printing Next-Gen. Robots Intelligent
Powertools Top floor Shopfloor End-2-End Digital Engineering
Sales/Marketing & Business Models Product Customization (5)
Product Usage Data (3) Batch-Size One (7) Work Environment
IoT-Enabled Manufacturing CPS De-Coupling, Product Memory
Servitization (4) (9) (6) (8) IoT Service Implementation Embedded |
Cloud (10) IoT Service Operation Adaptive Logistics Aftermarket
Services Remote Monitoring, Predictive Maint. (12) Source:
www.enterprise-iot.org