This research effort seeks to develop more accurate aircraft maintenance requirements forecasts and repair package optimization techniques. The capability to forecast using meta-data generation and data mining techniques applied to the Air Force REMIS maintenance data collection system, along with data fusion from other sources, such as prognostics data, is evaluated. The feasibility of developing more meaningful and information rich meta-data from raw REMIS data is investigated along with data mining techniques and tools for applying time series statistical analysis techniques for identifying patterns and trends, either from raw REMIS data or from developed meta-data. The application of specific techniques such as data clustering, association, regression analysis, autocorrelation and data visualization are evaluated for their application to REMIS data using a variety of readily available open-source software packages geared toward these types of problems. The feasibility of leveraging and customizing selected open-source data mining related software solutions are evaluated for application to the aircraft maintenance forecasting problem and future commercialization.
Benefit: The anticipated benefit of this research is more efficient aircraft depot maintenance planning, which can also be expected to lead to higher aircraft mission readiness/availability and lower life cycle cost. This is expected to occur through improved meta-data design, data fusion, data mining and time series trending analysis techniques. Potential commercial applications include many organizations that maintain fleets of vehicles (aircraft, cars, trucks, etc.), which have raw historical maintenance data that can be analyzed to identify patterns and trends, leading to more optimized maintenance planning and execution
Keywords: Aircraft Maintenance, Maintenance Planning, Optimization, Data Mining, Forecasting, Support Cost
