This study aimed to investigate the feasibility of using a model based on particle swarm optimization (PSO) and support vector machine (SVM) to predict the unconfined compressive strength (UCS) of cemented paste backfill (CTB). The dataset was built based on the experimental UCS values. Results revealed that the categorized randomly segmentation was a suitable approach to establish the training set. The PSO performed well in the SVM hyperparameters tuning; the optimal hyperparameters for the SVM to predict the UCS of CTB in this study were = 71.923, ε = 0.0625, and γ = 0.195. The established model showed a high accuracy and efficiency on the prediction work. The R value was 0.97 and the MSE value was 0.0044. It was concluded that the model was feasible to predict the UCS of CTB with high accuracy and efficiency. In the future, the accuracy and robustness of the prediction model will be further improved as the size of the dataset continues to grow.

Download full-text PDF

Source
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC8952849PMC
http://dx.doi.org/10.3390/ma15062128DOI Listing

Publication Analysis

Top Keywords

support vector
8
vector machine
8
particle swarm
8
swarm optimization
8
svm predict
8
predict ucs
8
ucs ctb
8
high accuracy
8
accuracy efficiency
8
model
5

Similar Publications

Want AI Summaries of new PubMed Abstracts delivered to your In-box?

Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!