Designing a Conceptual Model and Data Extraction Pipeline for Flight Test Cards

Abstract

The goal of this bachelor's thesis is to design a conceptual model and a data extraction pipeline for flight test cards, which are protocols created as Word files and filled with results and remarks by the flight test engineer. The concept is necessary because, in the current DLR workflow, flight test cards are stored as PDF files, which is not an optimal solution for matching and comparing flight test card data. This creates the need for a database to ensure suitable accessibility and efficient retrieval of flight test card data. The solution is tailored to the needs of DLR, the German aerospace research and technology center, particularly for all institutes involved in flight test analysis. This is because DLR is actively developing and researching new technologies in the field of aeronautics, and understanding these flight tests requires data from flight test cards. To achieve this, various approaches for extracting data from flight test cards will be explored, including different OCR techniques and table detection methods. The most suitable approach was selected and implemented into a pipeline to automatically extract data from the flight test cards. For efficient storage and querying of the extracted data, a data model was designed to link the flight test card information with the recorded sensor data and the additional information of the flight test, enabling better analysis of the maneuvers performed at DLR. Therefore, the content of the flight test cards was analyzed, along with methods for storing this data, to develop a suitable database model that meets the specific requirements derived from the flight test card content. This addresses the need for flight test cards to fully understand the executed maneuvers. The results of this thesis demonstrate that it is possible to automatically extract data from flight test cards, offering a clear advantage over the current solution. However, as a conceptual model, it also highlights several areas for improvement before the system can be deployed in production. This thesis serves as a foundation for further development and enhancements, with the ultimate goal of integrating the system into DLR’s operational workflow for flight test cards.