Kandiri, Amirreza and Shakor, Pshtiwan and Kurda, Rawaz and Farouk Deifalla, Ahmed (2022) Modified Artificial Neural Networks and Support Vector Regression to Predict Lateral Pressure Exerted by Fresh Concrete on Formwork. International Journal of Concrete Structures and Materials, 16. ISSN 1976-0485
Text (Research Article)
s40069-022-00554-4.pdf - Published Version Download (4MB) |
Abstract
In this study, a modifed Artifcial Neural Network (ANN) and Support Vector Regression (SVR) with three diferent optimization algorithms (Genetic, Salp Swarm and Grasshopper) were used to establish an accurate and easy-to-use module to predict the lateral pressure exerted by fresh concrete on formwork based on three main inputs, namely mix proportions (cement content, w/c, coarse aggregates, fne aggregates and admixture agent), casting rate, and height of specimens. The data have been obtained from 30 previously piloted experimental studies (resulted 113 samples). Achieved results for the model including all the input data provide the most excellent prediction of the exerted lateral pressure. Additionally, having diferent magnitudes of powder volume, aggregate volume and fuid content in the mix exposes diferent rising and descending in the lateral pressure outcomes. The results indicate that each model has its own advantages and disadvantages; however, the root mean square error values of the SVR models are lower than that of the ANN model. Additionally, the proposed models have been validated and all of them can accurately predict the lateral pressure of fresh concrete on the panel of the formwork.
Item Type: | Article |
---|---|
Uncontrolled Keywords: | ANN, SVR, machine learning, lateral pressure, mix proportion, concrete, formwork |
Subjects: | Engineering > Computer engineering Engineering > Mechatronics engineering and machinery Engineering > TK Electrical engineering |
Depositing User: | ePrints deposit |
Date Deposited: | 04 Sep 2023 08:06 |
Last Modified: | 04 Sep 2023 08:06 |
URI: | http://eprints.tiu.edu.iq/id/eprint/1150 |
Actions (login required)
View Item |