Compressive Strength Prediction of Green Ultra-High Performance Concrete (GUHPC) Based on Adaptive Network-Based Fuzzy Inference System

Abstract

Green ultra-high performance concrete (GUHPC) is considered to be a new generation of construction materialsthat adapt to sustainable development and is gradually being used in the fields of bridge reinforcement, housefacades, and paving.To improve the efficiency of green ultra-high performance concrete in the experimental stageand to save the component material, the prediction of the 28-day compressive strength of green ultra-highperformance concrete has become a challenging task. According to the published literature, the compressivestrength of concrete is closely related to the material composition such as cement, fly ash, silica fume, sand, etc. Soin this study, 175 groups of related data of GUHPC were collected to form a database, and an artificial neuralnetwork system combined with IF-THEN fuzzy rules was utilized to establish a model that could better predict the28-day compressive strength of GUHPC. Three evaluation indicators, RMSE, R2, and MAPE, indicate that theprediction of the compressive strength of green ultra-high performance concrete made by the model is completelyreliable. Overall,this study successfully proposes a fuzzy artificial neural network model for predicting the 28-daycompressive strength of GUHPC, which provides a viable prediction tool for GUHPC in the experimental stage.