U.S. Air Force Research Lab Summer Faculty Fellowship Program

U.S. Air Force Research Lab Summer Faculty Fellowship Program

U.S. Air Force Research Lab Summer Faculty Fellowship Program

412th Test Wing (Edwards Air Force Base, California )

SF.60.23.B10127: Statistical Modeling

Poulson, Robert - (661) 277-4031

As part of the Air Force Test Center, the 412th Test Wing located at Edwards AFB plans, conducts, analyzes, and reports on all flight and ground testing of aircraft, weapons systems, software and components as well as modeling and simulation for the USAF. Statistically defensible testing is an inherent component of developing and conducting quality tests and is infused throughout the testing and reporting process. Opportunities for current and future research and development of statistical methods include the following topics:

Develop experimental designs to optimize flight time and minimize constraint on randomization.
Use Bayesian methods to build predictive models and develop posterior distributions of complex, hierarchical flight test systems.
Research tail sampling to better understand inferences on percentiles including order and extreme value statistics.
Develop methods for modeling time series and spatial statistics related to flight tests.
Develop categorical data analysis methods to improve testing of human interaction with flight test systems. Such as survey analysis, testing of audio and communication systems, and human interface with flight systems.
Build supporting software in R and/or Python packages that would facilitate implementation of statistical methods into flight testing.

SF.60.22.B10094: Application of Machine Learning/Artificial Intelligence to aircraft-related Communications

Brownlow, James - 661-277-4843

Aircraft verbal communications are frequently noisy and/or incomplete. This research effort is to develop an AI/ML approach to clarify / complete verbal or message transmissions. This entails prediction of missing words prediction of correct words for incomplete or 'non' words and characterization of error rates in aircraft communications

SF.60.22.B10070: Application of ML / Bayesian techniques to Reliability analyses (system growth models, software interrupts).

Brownlow, James - 661-277-4843

Complex Aircraft systems require reliability and maintainability analyses that support frequent aircraft-system updates, reliability growth and the analysis of software-induced failures. Interest is in robust analysis/modeling techniques / procedures that support these objectives.

SF.60.22.B10069: Develop an AI model based on large data sets (>100000 observations, as many as 100 variables) to be used to predict aircraft loads in flight.”

Brownlow, James - 661-277-4843

The advent of large data sets and digital engineering opens the door to the use of Machine Learning/Artificial Intelligence in modeling and analyses of flight test data. This investigation is designed to develop analysis and models from large data sets. Interest is in models developed for both real-time and flight-test analysis applications. Such models form part of the digital engineering efforts. Dimension-reducing, Bayesian, multiple time series are some examples of approaches that may be used.

SF.60.21.B0003: Machine Learning Tool for EW Analysis

Upperman, Gary - (661) 277-1464

The 771st Test Squadron performs developmental test on electronic warfare (EW) systems. We have been developing a machine learning (ML) tool in Python to assist with data analysis. The goal is to be able to train a model that will perform "site-matching" (data association/reconciliation between a radar warning receiver and truth data from an electronic combat range), and then use that model to automatically perform this analysis step on new data, accelerating the data reduction process.
Preliminary testing demonstrated great success, but further testing with more complex data sets has yielded mixed results. Our desired research program would investigate new ML algorithms that would be more successful in EW data association. Currently the algorithms used have been limited to the scikit-learn python package. We also would like to investigate using unsupervised machine learning to perform clustering on the data and attempt to identify previously unrecognized RF signals. The goal is to have an algorithm that will predict new data-sets with 95% accuracy as compared to our traditional manual analysis techniques.

SF.60.17.B0001: Flight Test Education and Test Techniques

Cotting, Chris - 661-277-4517

The UASF Test Pilot School is the world’s premier institution for flight test education. We teach flight test techniques and flight test planning methods for both piloted and unpiloted fixed wing vehicles. We are seeking highly-qualified and motivated individuals for collaboration in the advancement of current state-of-the-art methodology in specific research and teaching areas while enhancing the capabilities of tools employed by USAF TPS to support test and evaluation education. Opportunities for focused research include the following topics:

• Advancement of our ability to teach complex topics to our students in minimal time using modern engineering education techniques, including the flipped classroom, direct application of theoretical concepts to student flight tests, simulation exercises that reinforce academic concepts, and educational tools for use in classroom and homework exercises.

• UAS curriculum evaluation and development to include specific flight test techniques for UAS platforms. USAF TPS is building a UAS flight testing curriculum and welcomes input for review and evaluation from others. TPS is also researching techniques to increase pilot/operator effectiveness in the presence of multiple seconds of time delay between a ground station and aircraft. TPS is also interested in instrumentation of Group I, II, and III UAS.

• Human interaction with complex systems to include Handling Qualities of complex control systems to include adaptive/robust systems and the prediction of new operational flight envelopes when robust control is actively accounting for uncertainties.

TPS has state of the art flight simulation capabilities and rapid prototyping of control going from simulation model to flight within the same day. TPS has unique access to multiple aircraft, and staff capable of conducting advanced/complex flight testing.

Air Force Test Center

Medina, Kevin
AFTC Technology Transfer & Research Manager
1 South Rosamond Blvd, Bldg 1
Edwards AFB, California 93524
Telephone: 661-277-9111
Email: kevin.medina.1@us.af.mil

Test, Test

1 Research Ct.
Rockville, Maryland 20850
Telephone: 111-111-1111
Email: NewSFFPLabDirector@gmail.com