Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
A Comparative Study to Evaluate the Performance of Various Deep Learning Models for Damage Identification in a Cantilever Beam
Damage detection is a key point in maintaining the service life of any structure. A structure encounters several types of load throughout its lifetime; this results in different types of damage, starting from the critical region and propagating further to other parts. This study presents a comparative study of different types of deep learning models, namely, artificial neural network (ANN), autoencoder neural network, and one-dimensional convolutional neural network (1D-CNN) for the classification and quantification of damages present in a modelled cantilever beam. To perform numerical analysis, damage is introduced by reducing the stiffness of the cantilever beam. The study utilizes the frequency response function (FRF) as a damage-sensitive parameter and evaluates the models using metrics such as precision, recall, and confusion matrix. The 1D-CNN model demonstrates superior performance, achieving a maximum accuracy of 99.06% in training and 99.03% in validation. Moreover, the classification metrics reveal exemplary performance for all three models, with an F1-score of 1.
A Comparative Study to Evaluate the Performance of Various Deep Learning Models for Damage Identification in a Cantilever Beam
Damage detection is a key point in maintaining the service life of any structure. A structure encounters several types of load throughout its lifetime; this results in different types of damage, starting from the critical region and propagating further to other parts. This study presents a comparative study of different types of deep learning models, namely, artificial neural network (ANN), autoencoder neural network, and one-dimensional convolutional neural network (1D-CNN) for the classification and quantification of damages present in a modelled cantilever beam. To perform numerical analysis, damage is introduced by reducing the stiffness of the cantilever beam. The study utilizes the frequency response function (FRF) as a damage-sensitive parameter and evaluates the models using metrics such as precision, recall, and confusion matrix. The 1D-CNN model demonstrates superior performance, achieving a maximum accuracy of 99.06% in training and 99.03% in validation. Moreover, the classification metrics reveal exemplary performance for all three models, with an F1-score of 1.
A Comparative Study to Evaluate the Performance of Various Deep Learning Models for Damage Identification in a Cantilever Beam
Lecture Notes in Civil Engineering
Goel, Manmohan Dass (Herausgeber:in) / Vyavahare, Arvind Y. (Herausgeber:in) / Khatri, Ashish P. (Herausgeber:in) / Baniya, Surendra (Autor:in) / Singh, Saurabh Kumar (Autor:in) / Maity, Damodar (Autor:in)
Structural Engineering Convention ; 2023 ; Nagpur, India
26.10.2024
12 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Damage Identification in a Cantilever Beam Using Regression and Machine Learning Models
| Springer Verlag | 2024
Damage Identification in a Cantilever Beam Using Regression and Machine Learning Models
| Springer Verlag | 2024
Damage Identification of CFRP Laminated Cantilever Beam by Using Neural Network
| British Library Online Contents | 1998
Cantilever beam fixing structure, cantilever system and mounting method of cantilever beam
| Europäisches Patentamt | 2020