The Use of GA and PSO in Evaluating the Shear Strength of Steel Fiber Reinforced Concrete Beams

Shear failure in concrete specimens and beams is one of the most sensitive cases in construction projects. Since the shear strength of specimens is affected by various parameters, it is important to study them to arrive at a predictive relationship or model. This research has used the computational methods of machine learning to achieve a new evaluation of the shear strength of reinforced concrete (RC) beams. In the first step, 85 laboratory samples were examined. These samples had different geometric and non-geometric properties. The effect of each parameter was examined for smart models, then the optimal state of the parameters was determined. By optimizing the number of parameters, basic models were developed. The different structures of these models were investigated for the accurate evaluation of the shear strength of the samples. Finally, in the second step, the particle swarm optimization (PSO) and genetic algorithm (GA) were used to perpetuate and optimize the weights of the base models. The hybrid models were developed and applied with different structures for evaluation of shear strength and the best models were selected from them. The results showed that both models offer better performance than the base model. However, the GA-based model achieved higher accuracy and less error than the PSO-based model. Finally, the performance of these models was used for new data to evaluate their performance. As the main achievement of this research, these methods can be recommended and implemented for other conditions as well.