Tunned fuzzy system development to assess the initial bond stress of steel reinforced concrete structures

Tunned fuzzy system development to assess the initial bond stress of steel reinforced concrete structures

연구 분야: Software Development

학회: Signal, Image and Video Processing

The initial-slip stress (τs) behavior between profiled steel and concrete is a crucial component in steel reinforced concrete (SRC) composite structures. Due to the development of restricted test data sets, the current models utilized to predict the τs of profiled steel–concrete may lack reliability when applied at a larger scale. This study employs multiple machine learning approaches to provide a data-driven estimation of the τs of profiled steel–concrete. This study compares two hybrid optimal approaches, namely the Brown Bear Algorithm (BBA) and the Beluga Whale Algorithm (BWA), merged with a conventional machine learning methodology known as the Adaptive Neuro-Fuzzy Inference System (ANFIS). Eight input parameters related to the initial slip bond stress of profiled steel–concrete are identified using a computational database containing 177 test outcomes from earlier research papers. More precisely, the process of creating and evaluating the suggested framework involved using 75% of the data as a training set and the remaining 25% as a test set. According to the results of this study, the BBA-ANFIS model produces more reliable and robust outcomes compared to the BWA-ANFIS model, as well as XGB and AdaBoost from earlier literature, as indicated by the higher R2 value. The BWA-ANFIS technique demonstrated a high level of procedural dependability, with R2 values of 0.9854 and 0.9851, respectively, during the process's training and assessment stages. It is shown that the BBA-ANFIS performs better and is more dependable than the other models after comparing them and compiling all the data and outcomes.