Development of a New Modified Sonar Inspired Optimization based on Machine Learning Methods for Evaluating Compressive of High-Performance Concrete

Document Type : Regular Paper

Authors

1 Associate Professor, Faculty of Engineering, University of Science and Culture, Tehran, Iran

2 Ph.D. Student, Department of Civil Engineering, University of Science and Culture, Tehran, Iran

3 Ph.D. Student, Department of Biomedical Engineering, Northwestern University, United States

10.22075/jrce.2024.33564.2025

Abstract

The nonlinearity observed in high-performance concrete (HPC) can be attributed to its distinctive features. This study examines the effectiveness of expert frameworks in determining compressive strength, aiming to enhance accuracy through the development of a master artificial neural network (ANN) system utilizing the sonar inspired optimization (SIO) algorithm. The ANN model employs exploratory data to establish initial optimal weights and biases, thereby improving precision. Comparison with previous studies validates the accuracy of the proposed system, demonstrating that the SIO-ANN hybrid model offers finer estimation of high-performance concrete properties. Results consistently show a coefficient of determination (R2) exceeding 0.972 and a 50%-67% reduction in error rates compared to conventional fitting curve approaches. Parameters such as population, weight, and bias within the SIO-ANN framework are continuously updated and optimized to achieve optimal values efficiently. Additionally, the SIO-ANN model exhibits superior runtime performance compared to other models. Consequently, the proposed SIO-ANN approach emerges as a viable alternative for accurately assessing and predicting the compressive strength of high-performance concrete.

Graphical Abstract

Development of a New Modified Sonar Inspired Optimization based on Machine Learning Methods for Evaluating Compressive of High-Performance Concrete

Highlights

  • The effect of different parameters on HPC was investigated by linear and non-linear methods.
  • The performance of the ANN base model was improved by using the SIO algorithm.
  • A comparison was made between the results of the new hybrid model and the previous models.

Keywords

Main Subjects

Files

History

  • Receive Date: 16 March 2024
  • Revise Date: 17 April 2024
  • Accept Date: 29 April 2024

Share

How to cite

Statistics