Application of novel hybrid machine learning approach for estimation of ultimate bond strength between ultra-high performance concrete and reinforced bar

There are few studies on how ultra-high-performance concrete UHPCs interact with reinforced bar, and there is less data on non-proprietary UHPCs. On the other hand, experimental testing (pull-out tests), which determine the feature of the reinforced bars in UHPC, requires more time, more cost, and high error percentage in the results. Therefore, the purpose of this research is to assess the variables that affect the ultimate bond strength (UBS) between UHPC and deformed reinforcing bars in order to generate guidelines for the novel connection of field-cast UHPC. To forecast the UBS between UHPC and reinforced bars, enhanced machine learning (ML) models, such as support vector regression (SVR), artificial neural networks (ANN), and an adaptive neuro-fuzzy approach (ANFIS) have been developed, using the root mean squared error (RMSE), correlation coefficient (r), and coefficient of determination (R2). Referring and synthesizing the feature parameter selection, concrete compressive strength(fc), tensile strength, bond length, water/cement ratio, reinforcing bar strength, and bar diameter (L/D) are used as input to machine learning models derived from pull out test. The R2 values for ANN, SVR, and ANFIS were 0.8825, 0.9351, and 0.9848, respectively. The RMSE value for ANN, SVR and ANFIS were 0.8779, 0.8555, 0.7667, respectively, all representing the best performance ANFIS in this analysis, followed by SVR and ANN as the weakest analysis. The numerical relevance of several components from three models demonstrates that the proportion of embedded depth to reinforcing bar diameter has a considerable influence on UHPC bond strength that is consistent with experimental results.