A new model of artificial intelligence predicts the service life of airplane engines
September 19, 2013
Researchers from the University of Oviedo, in collaboration with Rolls Royce, develop a model capable of determining the tendency of deterioration of the system
Optimizing the security levels and the use of fuel are the main challenges faced by the departments of R&D&i of the most important aviation companiess. Researchers from the University of Oviedo, in collaboration with Rolls Royce, are developing a new system capable of determining with a high degree of reliability the tendency of internal deterioration of airplane engines. The information allows for the detection of potential errors and contributes to making the upkeep of the planes more efficient and profitable.
Researchers from the group of Metrology and Models of the Department of IT have employed innovative data engineering tactics to design an intelligent system capable of diagnosing the possible tendency of deterioration of an engine based on different measurements. Professors Luciano Sánchez and Inés Couso and PhD student Álvaro Martínez have created models based on fuzzy logic, whose algorythms have already been successfully tested on the fleets of Rolls Royce.
The first results of the project have been successfully tested in a fleet and have received two international awards
The first results of the research conducted by the Asturian IT and mathematical engineers have been recently offered in the Fuzz-IEEE 2013 conference, which took place in India and where they received the award to the best conference. Furthermore, Rolls Royce has just named the work of this research team as the best innovation of 2013, granting them the 2013 Innovation Award for Publications. The company has selected the project among all the other researches conducted by its department of R&D in collaboration with other European universities.
The new system of artificial intelligence performs the prediction of the service life of an engine after calculation the deviation that exists between the fly data gathered by the different measurements and the expected values. This information allows it to know the exact maintenance needed by each engine and minimize the chance of unexpected situations. The capacity to anticipate the problem also noticeably reduces the direct and indirect upkeep costs.
Current maintenance systems for airplanes gather large quantities of flying data, measurements of the external conditions and also the working requirements demanded to calculate the condition of an engine. These results are contrasted with the average deterioration of the fleet to identify possible faults. The hypothesis on which the team of Metrology and Models have based their research aims at making the most of the current information given by the incomplete and low-quality measurements in order to predict the service life of an engine and what its tendency of deterioration will be.
The first tests conducted on airplanes show that high percentage of reliability of the new detection system. The researchers predict that they will finish their work throughout the first three months of 2014, a period during which the PhD Thesis associated with this project will probably be defended.