Data Refinery for Aviation Sustainment (DRAS)

The effectiveness and utility of algorithms and models used by the Army in support of Condition Based Maintenance (CBM+) are limited by noise in the source data. Tools and methods need to be developed and applied to reduce the noise in the source data. The solution needs to provide clean, relevant, and consistent data that increases the timeliness and effectiveness of the CBM+ analytical tools, algorithms, and models. The Data Refinery for Aviation Sustainment (DRAS) applies methods and techniques from statistics, data science, and data management to refine the source data prior to its integration into algorithms and models. DRAS characterizes the causes and types of noise in the source data, and then applies the appropriate methods to effectively filter this noise from the data while maintaining the integrity of the data itself. Noise in the CBM source data can be caused by factors ranging from erroneous sensor data to simple data entry or transcription errors. DRAS automatically detects and removes errors and inconsistencies in the source data, and identifies and applies methods to optimize data generated from multiple sources including small sample sizes.