Estimating the fatigue behavior of a family of aluminum conductor cables using ANNs applying the Poffenberger-Swart formula. (June 2022)

• Record Type: Journal Article
• Authors: Câmara, Eduardo César Bezerra; Kalombo, R.B.; Ferreira, J.L.A.; Araújo, José Alexander; Freire Júnior, Raimundo Carlos Silverio
• Is Part Of: International journal of fatigue. Volume 159(2022)
• Publication Date: 2022-06
• Subjects: Artificial neural networks -- Poffenberger-Swart -- Aluminum conductor cables -- Specific weight; Materials -- Fatigue -- Periodicals; Materials -- Fatigue; Periodicals; 620.1122
• DOI: 10.1016/j.ijfatigue.2022.106766
• Languages: English
• ISSNs: 0142-1123

Abstract

Highlights: A ANN architectures developed capable of estimating the fatigue behavior of aluminum overhead conductors, using Poffenberger-Swart formula. The use of the Poffenberger-Swart formula allows the use of various characteristics of the cable in ANN with a simplified way. Nevertheless, the linear weight of the rope is important information to be able to model the fatigue behavior of these structural elements. Abstract: The aim of this article was to develop an artificial neural network (ANN) architecture capable of estimating the service life of a family of aluminum conductor cables, using the K constant from the Poffenberger-Swart formula as reference, and specific weight ( W ). The database used to train and test the neural architectures was obtained from fatigue tests conducted at the cable laboratory of the Fatigue, Fracture and Materials Group (GFFM) of the Federal University of Brasilia (UnB). The ANNs were used to construct constant life curves for these conductor cables and the results compared with the values obtained for other ANN models presented in earlier studies. In addition to reducing the complexity of the ANN architecture, the results show that incorporating the Poffenberger-Swart formula to the ANN model also decreases the error obtained between the ANN and the values used in training and testing. The functions code produced in this paper will be provided in open-source format (https://github.com/raimundo-freire-junior/AL-CONDUCT-CABLES ).