Flightscope: Flight Analysis Designed for Big Data
Client:Naval Surface Warfare Center Carderock Division
- Analyze the enormouse volume of flight data collected from naval aircraft
- Improve the ability to deliver comprehensive information in order to create a flight data analysis and visualization framwork, dynamic database, model adaptation, and automated self-monitoring algorithms
- Develop three key technologies: the adaptive sensor, self-monitoring capabilities and segmentation/ flight-print methodologies
- Provided an increased awareness and an objective basis for intervention, such as specific pilot training or changes in aircraft maintenance practices
- Resulted in a highly robust analysis approach specifically designed to address the challenges of big data
Flightscope was designed to help the Naval Surface Warfare Center Carderock Division (NSWCCD) support fleet operations by enabling automated techniques to analyze the enormous volume of flight data collected from Naval aircraft, with sufficient detail to detect leading indicators of potential mishaps in a timely manner. CSC in close partnership with its client helped create a powerful flight data analysis and visualization framework, dynamic database, model adaptation, and automated self-monitoring algorithms. Unlike previous efforts, Flightscope is designed to work with all available data, vastly improving its ability to deliver comprehensive information. The information derived using Flightscope provides increased awareness and an objective basis for intervention, such as specific pilot training or changes in aircraft maintenance practices.
A defining aspect of Flightscope was that it be able to analyze all the data collected from recorder systems on fixed-wing and rotary-wing aircraft. This amounts to terabytes of data describing descents, landings, autorotation (helicopters) and other complex maneuvers collected over several years of flights. Analyzing such a large volume of data presents many challenges: the ability to consider all possible combinations of the data (scenarios); the need for automation and high reliability; and improved communication between user and machine.
Adaptive Sensor and Other Innovations
To address these challenges, the CSC team working with their Navy client developed three key technologies for Flightscope; the adaptive sensor, self-monitoring capabilities and segmentation/flight-print methodologies.
The adaptive sensor emulates the capabilities of a human analyst. It enables autonomous, undisruptive adaptation of the knowledge base and data models. This is done by modeling the human reasoning process and storing new content so it is independent from previous knowledge. For example, the adaptive sensor can recognize distinct and anomalous operations and store their data without disrupting the original data structure. This enables automated identification of new conditions, removing the element of chance in finding them and preventing potentially dangerous conditions from going unnoticed. The adaptive sensor has been awarded a U.S. patent.
The self-monitoring technology injects reliability into the analysis. Self-monitoring provides a fail-safe mechanism that can identify when a process in the system makes a mistake. People rely on self-monitoring all the time; automated procedures handling big data must do the same.
Segmentation/flight-print methodologies are data reduction and representation strategies that reduce data to a form that allows portability of the algorithms between applications, and autonomous manipulation of the data. The flight-print represents time-series data in a standardized format, which is used to produce a single image summarizing the information pertaining to an individual flight. The flight-print can also be constructed from multiple perspectives to address questions for different analyses.
Addressing the Challenges of Big Data
Taken together, these technologies result in a highly robust analysis approach specifically designed to address the challenges of big data. Flightscope can load and analyze several thousand files per day, a notable improvement compared to earlier efforts. It provides analysis for individual flights as well as an entire fleet.
Flightscope can be tailored for other aerospace applications and the core technology can be applied to big data analysis problems across industries.