Dynamically Reconfigurable Data Architectures for Aircraft Data Analysis and Anomaly Detection

In the field of aircraft health management, the integration and availability of historical data from multiple data sources is necessary for diagnostic and prognostic algorithms to realize their full potential. Especially important is the integration of historical in-flight operational data with ground-based maintenance data. Effectively integrating these data sources is difficult because key ground based maintenance fields are commonly entered in the form of natural language text rather than in a controlled vocabulary. Furthermore, with the ever increasing number of in-flight parameters available, selection of informative parameters for data mining is a time consuming, iterative process that requires domain knowledge. To address these difficulties in developing aircraft health management algorithms, Referentia Systems, Inc proposes to develop interactive software tools to mitigate these challenges. One software tool will perform information extraction from maintenance logs using a customized vocabulary. The structured results will then be integrated into a larger data warehouse. The second tool will enable interactive GUI-based development and execution of preprocessing and data mining algorithms for in-flight data. Results from each of these tools will be integrated into an existing dynamically reconfigurable health management data warehouse.