Aircraft Maintenance: Using Natural Language Processing (NLP) to identify recurring defects
Having the ability to identify potential recurring defects on an aircraft can greatly contribute to increasing the overall fleet availability of any airline. However, the ability to identify such recurring defects on individual aircraft is also dependent on several variables in the aircraft defect registration process. Every pilot describes a complaint differently and every maintenance engineer might use different wording to describe the same issue. Legacy systems tend to address this issue by working on the bases of (sub)ATA chapter usage, however this makes identification of potential recurring defects highly dependent on data quality.
With this upcoming feature in the AVILYTICS reliability module, we are using natural language processing to classify the actual complaint descriptions and actions being taken in order to determine if a certain complaint is reoccurring. In addition, by using metadata such as AMM references, component descriptions, part numbers and serial numbers the system is able to automatically link the potential recurring defect to the exact component installed and its position on the aircraft.
What is recurring defect analysis and why does it matter?
Recurring defect analysis revolves around analysing the technical complaints information of each individual aircraft in the fleet and determine if specific complaints keep reappearing on individual aircraft or across a certain fleet of the same aircraft type. Being able to identify such potential complaints can assist in preventing unscheduled maintenance downtime of aircraft and subsequently increase the overall aircraft /fleet availability.
Similar as to aircraft systems reliability monitoring, the information used for Recurring defect analysis are the actual technical complaints raised by flight crew (Pirep), maintenance engineers (Marep) or cabin attendants (cabin defects). The difference being that where system reliability monitoring looks at specific trend over time in aircraft systems, Recurring defect analysis looks at specific complaints themselves re-appearing over a given period on each individual aircraft. Such Recurring defects could be an identification of an impact by operational usage of the aircraft, weather conditions, negative effects of a modification or potential inadequate maintenance practices or component reliability.
Within AVILYTICS, the Recurring defect analysis allows to identify any potential repeating defects on either an individual aircraft or across multiple aircraft in the same fleet. With this enhancement of AVILYTICS ability to identify recurring defects, Maintenance Control Centers, Reliability Engineers, Technical Specialists and Troubleshooters alike will be able to focus their time on solving the actual causes of the recurring defect, rather than spend countless hours on analysing data in order to determine if a potential recurring defect case exists.
For more information and/or a software demo feel free to just leave your details below:
