Applying two optimization techniques in evaluating tensile strength of granitic samples

Abstract

The Brazilian tensile strength (BTS) test is the most common method to evaluate the tensile strength in mining and civil engineering projects. This paper aims to employ the genetic algorithm (GA) and particle swarm optimization (PSO) for predicting the BTS. For this work, linear and power equations were considered and their weights were optimized By GA and PSO. To achieve the objective of this research, a database including 80 sets of data were prepared so that dry density (DD), Schmidt hammer (Rn), and point load (IS50) parameters were used as the independent parameters. After modeling, the accuracy of PSO linear, PSO power, GA linear and GA power models were assessed using coefficient correlation (R2) and root mean square error (RMSE). According to the obtained results, it was found that both GA and PSO optimization algorithms proposed in this research predicted the BTS values satisfactorily; however, PSO power model with the R2 of 0.963 demonstrated a better generalization capability and it can be used for similar problems in the future. Also, the values of R2 for the PSO linear, GA Linear and GA power models were 0.958, 0.948 and 0.962, respectively.