Study on eccentric compression performance of carbon fiber reinforced concrete columns based on machine learning

The CFRP (Carbon Fiber Reinforced Polymer Composites) reinforced concrete columns have different characteristics, and the stress analysis under load is complex, which makes it difficult to establish a theoretical model. In recent years, machine learning has been widely applied in the field of civil engineering to solve nonlinear problems. This article first used experimental data from 91 groups of related literature to train four machine learning models, namely BPNN, POS-BPNN, SVM, and RFR, In the test set, BPNN, PSO-BPNN, SVM, RFR fitting precision were 0.8295, 0.9747, 0.9667, 0.7634, and it was found that machine learning can better predict the mechanical performance of CFRP reinforced concrete columns under eccentric compression. Then, the feasibility of four types of machine learning in predicting the eccentric compression performance of CFRP reinforced concrete columns was evaluated by evaluating the relationship between predicted and measured values. Finally, the main factors affecting the eccentric bearing capacity of CFRP reinforced concrete columns are ranked, and four models are used to predict the data in reference [4].