Enhancing the machine learning model with weighted features to predict the adhesion force between reinforced concrete and FRP materials

Abstract

This study is dedicated to developing a learning model for predicting the adhesion force between reinforced concrete and Fiber Reinforced Polymer (FRP) materials. The hybrid model, named Jellyfish Search Optimized Weighted Feature Stacking System (JS-WFSS), is constructed using a weighted stacking system optimized through jellyfish search. The analysis results clearly indicate that the JS-WFSS optimized hybrid model demonstrates a higher level of prediction accuracy compared to previous studies. Beyond model construction, this study assesses the significance of input variables in determining the bearing capacity between concrete materials and reinforced FRP material by scrutinizing the variable counterweights within the predictive model.

Keywords: Adhesion between FRP and concrete; FRP materials; machine learning model; optimization; jellyfish search optimizer.