Data-Driven Systems Engineering · Long training periods Requires understanding of model and...

21
Data-Driven Systems Engineering Turning MBSE into Industrial Reality SECESA 2018 – University of Strathclyde - Glasgow

Transcript of Data-Driven Systems Engineering · Long training periods Requires understanding of model and...

Page 1: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

Data-Driven Systems EngineeringTurning MBSE into Industrial Reality

SECESA 2018 – University of Strathclyde - Glasgow

Page 2: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

Outline

Industrial reality

Data-driven SE

Implementation

Conclusion

1.

2.

3.

4.

2

Page 3: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

1. Industrial reality

3

Page 4: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

Video link:https://youtu.be/f-SP-PsAXeI

4

Page 5: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

Most companies in the space industry

still use a document-based approach to

engineering.

Photo created by Pressfoto - Freepik.com 5

Page 6: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

86%of engineers’ time is spent on non-engineering work

100%*there are inconsistencies in the documentation

33%is spent on searching, reading and writing documentation

*extrapolation from personal experience 6

Page 7: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

ConsequencesFailuresMars Climate Orbiter1999

Loss of spacecraft due to ground-based computer software which produced output in non-SI units of pound-force seconds (lbf·s) instead of the SI units of newton-seconds (N·s).

7

Page 8: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

2. Data-driven systems engineering

8

Page 9: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

Good data management is

becoming more and more essential for

engineering companies in the

space industry.

Tony Stark in Iron Man9

Page 10: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

What about MBSE?

Identified problems

● Too complex and inflexible for practical use● Long training periods● Requires understanding of model and modeling

language● Much emphasis on the representation form of

models and overlooks the importance of the underlying data for the verification and analysis of models

10

Page 11: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

Data-Driven Systems Engineering (DDSE)

is a novel method, which enables a wider spread of MBSE throughout

the industry and allows for consistency of documentation

11

Page 12: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

DDSE

Our definition

A process where engineering data and associated

structure, links and connections constitute the

foundation of the systems engineering process.

12

Page 13: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

DDSE

Benefits

● Consistent database of connected engineeringvalues

● Automation● Traceability and transparency● Optimization possibilities

13

Page 14: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

3. Implementation

14

Page 15: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

To implement DDSE, it is necessary to start

by creating an infrastructure that

enables easy instrumentation and

data access for all stakeholders.

15

Page 16: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

The engineering data pyramid

Machine learning, Artificial Intelligence

Multidisciplinary design optimization

Budget, comparison chart, design structure matrix, history timeline

Search, filter, connection tracking, impact analysis, notifications

Self learn

Optimize

Aggregate / summarize

Explore

Structured storage

Collect / exchange

Component tree, matrices, margins, unit conversions

REST API, plugins and interfaces to specialized tools

*inspired by “The AI Hierarchy of Needs” by M. Rogati 16

Page 17: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

Tool connections

To be truly data-driven, it is important to

connect tools and systems from different

disciplines and areas through open APIs.

17

Page 18: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

Browser-based tool stack

Benefits• Data exchange

through standard APIs

• Automated toolinteractions and connections

• Creates a singlesource of ’truth’

18

Page 19: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

4. Conclusion

19

Page 20: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

DDSE is a proposed solution that aims to enable model-based

engineering on a practical level for space companies

throughout the project lifecycle.

20

Page 21: Data-Driven Systems Engineering · Long training periods Requires understanding of model and modeling ... Machine learning, Artificial Intelligence Multidisciplinary design optimization

Thank you!

Text me your email address to get the slides:

+351 964 211 963

https://api.whatsapp.com/send?phone=351964211963