Small Business Innovation Research/Small Business Tech Transfer
Distributed Data Mining for Aircraft Health Management
Project Description
NASA, DoD, and commercial aircraft operators need to transform vast amounts of aircraft data accumulated in distributed databases into actionable knowledge. We propose distributed algorithms for data-driven health monitoring on aircraft, aircraft fleet, and national airspace levels. The proposed algorithms are based on distributed optimization formulation, and, unlike existing distributed processing methods, have rigorous guarantees of producing the same results as centralized processing would do. Our algorithms will be implemented in an open scalable framework that allows integrating distributed data and federated third party algorithms for anomaly detection, diagnosis, prediction, and prognosis. We will apply the proposed approach to aircraft performance monitoring from FOQA data. We will train regression models of aircraft performance using distributed agents associated with different data sets, locations, and organizations. The trained models will be then used for anomaly detection, diagnosis (fault isolation), prognosis (forecasting), and mitigation (decision support). This project will develop web-based distributed open architecture software implementing the proposed optimization-based approaches and demonstrate scalability to at least 10 TB of data. Besides the developed algorithms, we will explore integration of third party algorithms into the distributed environment. The developed technologies will be applicable to a broad range of aircraft-related and other problems.
More »
Anticipated Benefits
*Airline FOQA.* The proposed demonstration of distributed FOQA data monitoring provides a direct path for transitioning the results of this SBIR projects into military and commercial aircraft fleets. The data-driven modeling would make advanced FOQA monitoring easy to implement. The proposed technology could allow the two airlines to share the data-driven performance models of their aircraft while processing their respective FOQA data privately. *Propulsion.* The fleet-wide data-driven monitoring technology developed in this SBIR can be applied to jet engine fleets. The technology developed in the proposed SBIR project would provide scalability above and beyond what is available in existing systems. *Smart grid.* This is an area of rapidly increasing and potentially immense societal and business impact. Monitoring of power distribution systems is a major application that is not addressed at present. The proposed distributed data-driven monitoring technology is of interest for this area. *Semiconductor manufacturing.* If undetected in time, a fault in a semiconductor manufacturing process tool could lead to losing hundreds of high-cost wafers passing through the tool. Being located in Silicon Valley we pursue specific opportunities in this area.
This SBIR topic supports objectives of NASA Aviation Safety Program by enabling improved health management. One of the goals of this project is to support transition of NASA's data mining research (both internal and external) into practical use in the aviation industry. NASA data mining algorithms that are formulated as optimization algorithms can be integrated into the proposed distributed optimization and software framework. In addition to being a platform for deploying algorithms at airlines, the proposed distributed framework can support NASA in building national aviation safety resources. Using the developed technology, the airlines, industry vendors, and government can share the mined global knowledge without actually sharing the distributed local data used in its computation. An additional potential use of the developed distributed data mining technology is for validation of aircraft software after its initial deployment. Validation requires extensive test coverage. Monitoring an aircraft fleet would cover a much broader range of conditions than monitoring any single aircraft in the fleet. More »
This SBIR topic supports objectives of NASA Aviation Safety Program by enabling improved health management. One of the goals of this project is to support transition of NASA's data mining research (both internal and external) into practical use in the aviation industry. NASA data mining algorithms that are formulated as optimization algorithms can be integrated into the proposed distributed optimization and software framework. In addition to being a platform for deploying algorithms at airlines, the proposed distributed framework can support NASA in building national aviation safety resources. Using the developed technology, the airlines, industry vendors, and government can share the mined global knowledge without actually sharing the distributed local data used in its computation. An additional potential use of the developed distributed data mining technology is for validation of aircraft software after its initial deployment. Validation requires extensive test coverage. Monitoring an aircraft fleet would cover a much broader range of conditions than monitoring any single aircraft in the fleet. More »
Primary U.S. Work Locations and Key Partners
| Organizations Performing Work | Role | Type | Location |
|---|---|---|---|
| Mitek Analytics LLC | Lead Organization | Industry | Palo Alto, California |
Ames Research Center
(ARC)
|
Supporting Organization | NASA Center | Moffett Field, California |
Primary U.S. Work Locations
-
California
Suggest an Edit
Recommend changes and additions to this project record.
This is a historic project that was completed before the creation of TechPort on October 1, 2012. Available data has been included. This record may contain less data than currently active projects.