AI models predict the need for aircraft maintenance

AI models can be used to make health predictions for aircraft systems. This makes it possible to tailor the maintenance process to these predictions and thus makes the maintenance more efficient.

This emerges from a large study conducted by several universities and industries under the name ReMAP. TU Delft is the project manager. The Dutch University calls the research and the developed AI models an important step in the modernization of aircraft maintenance.

Integrated Fleet Health Management

ReMAP is a Horizon 2020 funded project. The project started on June 1, 2018 and will end at the end of this month. ReMAP aims to contribute to Europe’s leading position in aviation by developing an open source aviation maintenance solution. This system is also known as the IFHM (Integrated Fleet Health Management) system. The system replaces maintenance at fixed periodic times with an adaptive maintenance model. With this model, the maintenance is performed when it is actually needed.

The following parties are involved in ReMAP:

  • Delft University of Technology (NL)
  • ATOS Spain SA (SP)
  • Cedrat Technologies (FR)
  • Collins Aerospace (IE)
  • Ecole National Supérieure d’Arts et Metiers (FR)
  • Embraer Portugal SA (PT)
  • Instituto Pedro Nunes (PT)
  • KLM Royal Dutch Airlines (NL)
  • National Office of Aerospace Studies and Research – ONERA (FR)
  • Optimal solutions (PT)
  • Smartec SA (CH)
  • University of Coimbra (PT)
  • University of Patras (GR)

From fixed maintenance steps to continuous health monitoring

Project Manager Bruno Santos, Assistant Professor Airline Operations, says in a news report: “We have succeeded in modeling the entire maintenance process for different aircraft fleets. This will in the future make it possible to transform current aircraft maintenance based on fixed time intervals, and maintenance due to defects , for continuous health monitoring of systems.Systems are then replaced exactly when needed, reducing waste.In addition, the team has modeled the complex process of maintenance planning, which is currently largely carried out by hand.It also takes into account changes and disruptions, so it approximates practice better than existing static models. This allows maintenance to be planned longer in advance. “

The potential savings on aircraft maintenance are huge. In Europe alone, the savings can be expected to reach 700 million euros annually, according to the Advisory Council for Aviation Research in Europe (ACARE).

Plan months ahead

Paul Chün, Vice President Technology Hub KLM Engineering & Maintenance: “With this concept, we can consider replacing the current manual maintenance scheduling, which is no more than a few weeks ahead, with this automated scheduling process that allows us to plan several months ahead. The benefit to travelers is clear: less unplanned maintenance results in less delays and cancellations of flights. “

Santos: “With our integrated ReMAP approach, we have made an important contribution to making optimal use of condition-based maintenance in the commercial aviation industry. We have also developed an open IT platform that allows AI developers to run their forecasting or scheduling algorithms based on actual operating data with a few clicks. This promotes the development of innovative third-party solutions to move rapidly from maintenance at regular intervals to a truly adaptive, state-of-the-art maintenance approach in the regulated aviation domain. “

Discover damage to composite aircraft structures

In addition to research into AI models, other studies are also part of ReMAP. One of these studies focuses on the possibilities of using diagnostic and prognostic models for composite aircraft structures. Manual inspections of these structures are time consuming. For example, damage to composite aircraft structures is often not visible on the surface. This makes it difficult to detect damage.

Within the ReMAP project, diagnostic and forecasting systems have been developed that use AI to map injuries, locate injuries and determine the degree of injury. This model has been tested for two years by the Faculty of Aerospace Technology at TU Delft and the University of Patras (GR). Based on this test period, a public knowledge base has been established.

Author: Wouter Hoeffnagel
Photo: Free images from Pixabay

Leave a Comment