Abstract

LEMACHE-CAIZA, Karina; GARCIA-MORA, Félix; VALVERDE-GONZALEZ, Vanessa  and  LOPEZ, Efraín Velastegui. The machine learning approach to industrial maintenance management. Universidad y Sociedad [online]. 2023, vol.15, n.3, pp. 628-637.  Epub June 30, 2023. ISSN 2218-3620.

Smart manufacturing and Industry 4.0 innovation worldwide are part of the technological transformation to create management systems and ways of doing business that optimize manufacturing processes, achieve greater flexibility and efficiency, and respond in a timely manner to the needs of their market. Machine learning is a technology that is able to reliably predict certain outcomes from a prepared model by training it with previous input data and its output behavior. The research carried out was aimed at comparing machine learning models for the detection of failures in twin-flow turbojets extracted from the NASA Prediction Centre of Excellence Repository. The results obtained are compared with real data to verify the accuracy resulting in the Random Forest algorithm as the best model run with normal and optimized parameters with an f1-score of 99.949% and 99.99% respectively. Finally, it is known that in the database it is not possible to perform a reliable and valid extraction of the main features by machine learning, due to its particularity in the operating conditions. It is also important to mention that the SVM model was not run with hyperparameters. It is advisable to use deep learning matching methods because of their accuracy in classifying the data and drastically reducing the computational load when running the model.

Keywords : Machine learning; Twin-flow turbojet; Industrial maintenance.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )