Analysis of ductility of hybrid fiber ultra-high performance concrete based on improved GA-BP neural network

Abstract

Ultra-high-performance concrete is a cement-based material with ultra-high strength, outstanding toughness, and excellent durability, which enables structures to achieve larger spans and lighter dimensions. In this paper, an improved GA-BP neural network model is constructed based on BP neural network, which is optimized and improved by the GA algorithm. Then, the experimental data were input into the improved GA-BP neural network model by designing experiments with different types and volume doping of blended fiber UHPC, and the ductility of blended fiber UHPC was analyzed in terms of compressive strength and tensile strength. In terms of compressive strength, the compressive strengths of each group of PE fibers with 0.5%, 1.0%, and 1.5% volume doping were from PD/S<PD/H<PA/S<PA/H. In terms of tensile strength, the 1.0% volume doping of short straight type S and 1.5% volume doping of end hook type H had the best effect, and the tensile strength reached 12.44 MPa. GA-BP neural network can effectively analyze the factors influencing the ductility of blended fiber ultra-high performance concrete.

Keywords

• hybrid fiber
• uhpc
• bp neural network
• ga algorithm
• ductility
• 68m01