INDI

As digitalisation progresses, the abundance of data and database systems for aircraft maintenance continues to increase. Methods from the fields of data mining and machine learning are suitable for maintaining the overview and evaluability of available information and also for increasing the efficiency of diagnoses during maintenance. The aim of the INDI project (“Intelligent Data Utilization in Maintenance”) is to develop an integrated data analysis framework that filters and links data from different databases in a meaningful way and enables intelligent analyses based on the principle of “Big data to smart data”. On the basis of this concentrated information, recommendations for action are to be generated on an automated basis, which should support aircraft maintenance in the areas of fault detection and isolation with corresponding new digital products. The work in the project is divided into the development of the diagnostic model (TU Darmstadt) as well as the provision of data and know-how from aircraft maintenance (Lufthansa Technik).

Approach

By concentrating and processing relevant information in order to be able to automatically identify component and system-specific error patterns, the TU Darmstadt develops and evaluates a data analysis framework to support aircraft maintenance. Methods from the field of clustering (e.g. k-Means, Principle Component Analysis) are considered for data fusion. Artificial intelligence methods with machine learning tools (e. g. Gaussian Process, Support Vector Machines, Neural Networks) are promising for the subsequent error diagnosis. Defined use cases are used to evaluate the diagnostic accuracy of the trained data analysis framework. A graphical user interface tailored to the requirements is to present the diagnostic results for system engineers and maintenance personnel in a suitable way and to issue appropriate recommendations for action.