Applying Informatics to Autonomy in Experimentation

Transcription

1 Properties Applying Informatics to Autonomy in Experimentation Process Structure 1/27/2015 Jason Poleski Michael Krein Lockheed Martin Advanced Technology Laboratories Lockheed Martin Proprietary Information 1

2 Lockheed Martin Corporate Alignment Aeronautics Information Systems & Global Solutions Missiles & Fire Control Mission Systems & Training Space Systems International Copyright 2015 Martin by Lockheed Martin Corporation Lockheed Proprietary Information 2

3 ATL s Role in Lockheed Martin Aeronautics Information Systems & Global Solutions Missiles & Fire Control Mission Systems & Training Space Systems International Copyright 2015 Martin by Lockheed Martin Corporation Lockheed Proprietary Information 3

4 Advanced Technology Laboratories Innovation with Purpose We partner with the Business Areas to engage the S&T community, develop innovative and disruptive technologies, and provide ground-breaking capabilities to extend programs of record and expand into new lines of business. Engaging the S&T customer to support creation of the next generation LRP Lockheed Martin Proprietary Information 4

5 The Materials Science Golden Triangle Properties Temperature Humidity Feedstock Catalysts Process Structure Lockheed Martin Proprietary Information 5

6 Integrated Computational Materials Science and Engineering (ICMSE) A Vision for Innovation in Future Systems Development Refine / Revise Material / Mission Capabilities Refine / Revise Material Design Tools Customer Market Need System / Platform / Mission Reqts. Material Reqts. Experts- on-the- Loop Models to Determine Processing Approach Linking Tools Models to Predict Properties OODA Loop Concurrent Material / Product Design Cycle Linking Tools Linking Tools Cycle to convergence Models to Predict Product Performance Tools to Predict Microstructure Tools to Simulate Full- Scale Testing Full-Scale Models Conduct Full Scale Tests Final Design / Firm Configuration Low Rate Build Production Lockheed Martin Proprietary Information 6

7 Autonomous Research: Short Circuiting Current Materials-to-Manufacturing Design Process Structure Design Select Material Qualify Material Verify Process Ass y, Intgn & Test Certify Parts LRIP Autonomous Research No Rapid explora,on of materials, process, and design choices allow early op,miza,on. Pass/fail designs early in the process. Reduce surprises in design manufacturability and performance. Lockheed Martin Proprietary Information 7

8 Lockheed Martin Proprietary Information 8

9 Informatics Methods Applied to Autonomous Research Lockheed Martin Proprietary Information 9

10 Informatics Tool Developed by LM The Material Data Mining, Modeling and Management (MD-M3) Tool. Analyzing trends in data sets, e.g. product performance and efficacy Determining inter-measurement relationships and dependencies, data trends/analysis Has over 10 analysis algorithms that run in series or in parallel to predict results based on input numerical data, the next set of experiments (configurations) Significant visualization techniques to provide the user with insights that are not clearly apparent Lockheed Martin Proprietary Information 10

11 Current Approach Integrate with ARES platform Create Operational Interface / GUI Develop optimized models Define an optimization strategy Process Constraints / Targets Database Process Data Document Document Process Optimization Genetic Algorithms Task Model Automated Experimentation and Characterization Next Experiments Lockheed Martin Proprietary Information 11

12 LM MD-M3 / AFRL ARES Integration: 1 st demonstrations of learning Graphic courtesy of Pavel Nikolaev - AFRL Autonomous Process-Structure-Property Navigation Lockheed Martin Proprietary Information 12

13 Next Steps Improve model realism Tighter integration with experimental platforms Process Constraints / Targets Database Process Data Document Document Process Multimodel Process Optimization Genetic Algorithms Task 1 Model Task 2 Model Task 3 Model Automated Experimentation and Characterization Next Experiments Lockheed Martin Proprietary Information 13

14 Thank you Thank you to my colleagues for their support: Michael Krein- Lockheed Martin ATL Shashishekar Adiga Lockheed Martin ATL Rick Barto Lockheed Martin ATL Benji Maruyama Air Force Research Laboratory Pavel Nikolaev Air Force Research Laboratory Daylond Hooper Air Force Research Laboratory Lockheed Martin Proprietary Information 14

15