The research of rebound in dry mixing shotcrete

In this thesis, the amount of material rebounding in the shotcrete application was obtained to be used in the dataset of Adaboost algorithm, which is one of the methods of community learning with the experimental application. In the study, as two of 14 samples were produced as plain and the others were produced by adding silica fume and fly ash that were replaced of the ratio of 10% and %20 of cement volume, and by adding polypropylene fiber at the rates of (5 kg / m3 and 10 kg / m3). During the application of the shotcrete to the panels, the rebound material was gathered and then weighed and registered as the data. According to the findings that were obtained from the experimental study; the lowest compressive strength of the samples for the 28th days was observed in the sample, which is 20% fly ash substitution, and the highest compressive strength was observed in the sample that has 10% silica fume substitution. When considering the results of the rebound, the most rebound was observed in the plain sample. The plain sample was followed by the samples with %20 fly ash substitution. The lowest values were observed with the silica fume substitution. Especially, a reducement of the rebound was detected with the increment of the amount of the silica fume. As a result of experimental studies, dependent and independent parameters were determined in the data set that was obtained. The Hyperparameter model and Boosting method that give the optimum results for the training of the model were determined. In conclusion, %84.25 success was achieved in the prediction performance of the model. In order to test the performance of the suggested model, It was compared the traditional machine learning algorithms with on the same data set. As a result, it has been found that the proposed model has the highest accuracy.