“There are no easy solutions. Collaboration of all stakeholders in the chain of aircraft maintenance is key.”Paulo Rupino, scientific coordinator from Instituto Pedro Nunes (IPN), explains the steps that have been taken and the challenges the researchers encountered
We know Condition Based Maintenance (CBM). What’s new?
Rupino: ”The idea of CBM is to replace parts exactly when they really need to be replaced. Sooner if needed. Later if possible. CBM nowadays is used in a very limited manner. In addition to what the data might say, there are always physical inspections and periodic checks by humans. What we want in the future is that we can trust solely on the data to make the decision about replacing or keeping components. There are data-driven initiatives such as Skywise from Airbus and AVIATAR from Lufthansa, but they are dependent on those companies. ReMAP will offer a company-independent open IT-platform. It will run Artificial Intelligence (AI) algorithms to infer the maximum useful life of aircraft components.”
Fleetdata is sensitive information. How does ReMAP handle this?
Rupino: “The more data, the better the AI-techniques can function. But we are talking about several airlines, that may be competitors, sharing that data. The IT platform architecture must be designed in such a way that airlines can collaborate without violating secrets and without exposing competitive information that is embedded in that data. Because collaboration is key. Five architectural scenarios came up after intensive workshops with the consortium, which consists of relevant European aviation stakeholders, and external companies. Every alternative has some pros and cons. Some are technically better, but are constrained by legislation or the decision of companies not to expose data. Having discussed all the scenarios, the consortium made an informed decision on how to proceed. Data will also be key for system validation. The consortium must show regulation authorities that the system actually works and that we can rely on it for the future of maintenance.”
Choice of ReMAP IT-architecture
The ReMAP IT platform will adopt a ‘distributed architecture’, which means data will never leave the airlines. The AI-tools will be trained by each participant using their own data from their fleet. That trained model will be shared on a central location, but not the data. Other airlines can select those trained models from this central location and use them with their own data. For instance, companies with Boeings 787 train the AI-tool using their own data. This AI-model will be shared. Other companies may run those models with their 787 data to see what they come up with and which model works better for their case.