Multiaxial Strength Criterion Model of Concrete Based on Random Forest

• DOI: 10.3390/math11010244
• Publication Date: 2023-01-04T07:15:42Z
• Authors: Xingqiao Chen; Dongjian Zheng; Yongtao Liu; Xin Wu; Haifeng Jiang; Jianchun Qiu
• Keywords: Basis (linear algebra)

The concrete strength criterion is the basis of analysis and evaluation under a complex stress state. In this paper, large number multiaxial tests were carried out, many mathematical expressions criteria proposed based on geometric characteristics assumption convex function. However, rationality function limits use these criteria. particular, misjudgment will occur near failure curve surface. Therefore, paper does not assume shape in advance. By collecting experimental data using machine learning method, it proposes method hidden Based 777 groups data, random forest (RF), back propagation neural network (BP) radial (RBF) models used to analyze verify feasibility effectiveness method. Subsequently, results compared with Ottosen criterion, Guo Wang Drucker–Prager (DP) criterion. show that consistency between model established by algorithm (especially forest) higher than multiaxis preset surface shape. Moreover, physical significance clearer, deficiency hypothesis avoided more universal.