Andromeda – The Future of Aviation Predictive Analytics
A Case Study
Background Context
In 2019, NLP Logix partnered with Andromeda Systems Inc. (ASI) the leader in Department of Defense and commercial supportability to enhance maintenance of the military’s F-35 fleet of the Lockheed Martin single-seat, single-engine, stealth multirole fighter aircraft. The program intended to optimize unscheduled maintenance of the F-35.
This project began as ASI focused on the next generation of an existing initiative called Artificial Intelligence Prognostic Steering™ (AIPS), designed to enhance high-end transportation machinery, such as the F-35 operation and performance.
The next evolution of AIPS was transformed by utilizing a combination of machine learning and statistical modeling and integrating these models into the F-35 maintenance workflows. Robert McCutcheon, Project Manager at ASI said, “AIPS leverages decades of supportability experience from ASI and the Machine Learning expertise from NLP Logix.”
The Gnarly Problem
ASI began development of an AI-powered aircraft maintenance application called AIPS. AIPS’s intention was to provide maintenance personnel with the optimal suggested actions to a given alert by on-board aircraft systems with the goal of reducing cost and maintenance time.
The ASI and NLP Logix Teams worked on two versions of the algorithm. Version 1.0 identified the optimal order of maintenance step implementation. Version 2.0 incorporated opportunistic maintenance, component age, weather, and environmental impacts.
Challenges Version 1.0:
- Flightline data is difficult to understand without context of having been around aircraft. NLP Logix relied on the partnership and expertise of the ASI Team to understand and connect data.
Challenges Version 2.0:
- External data sources needed to be incorporated. The understanding of operational impacts, environmental factors, and the fleet deployment schedules were required for the model to effectively identify the estimated remaining use life of aircraft parts.
Our Strategic Approach
The algorithm allowed for predictive analytics in forecasting based upon F-35 customer-defined metrics and an algorithm predicting the demand signal for supply from the flight line, later to be incorporated into the current F-35 AIPS module, Artificial Intelligence Prognostic Steering™ – AIPS – Andromeda Systems Incorporated (androsysinc.com).
Having identified ASI’s key points of interest, NLP focused on the following deliverables:
- An AIPS algorithm within the current F-35 AIPS module and with the capability to predict the demand signal for supply from flight line.
- An enhanced AIPS algorithm within the current F-35 AIPS module with the capability to perform predictive analytics in forecasting based upon F-35 customer-defined metrics (aviation operational impacts, environmental factors, fleet schedules, etc.)
Katie Bakewell, NLP Logix’s Modeling and Analytics Team Lead stated, “By leveraging python and SQL servers the combined NLP Logix and ASI teams were able to build a solution that provided the necessary information while meeting the requirements of government technology restrictions.” With this development McCutcheon said, “AIPS will optimize maintenance events reducing asset management costs and increasing asset availability.
Operational Success
Once the predictive module was developed and delivered general observations from the ASI Team were as follows:
- Maintenance times were notably reduced
- False Alarms (FA) were identified at a rate of 96%
- Significant unnecessary maintenance was avoided
- Realized Opportunistic Maintenance Events considerably reduced air vehicle downtime
Ted Willich, the CEO of NLP Logix, said the AI powered application could easily be configured to support other high-end complex machinery that requires preventive maintenance. McCutcheon said, “AIPS is applicable across any industry where maintenance data is collected at the point of performance. Cost, maintenance and down times will be dramatically reduced while asset availability will be greatly enhanced.”
Learn more:
