Research On Mechanical Property Prediction Of UHPC Based On Ensemble Learning

As society continues to develop,modern buildings have higher requirements for the performance of concrete,especially for ultra-high performance concrete(UHPC).However,traditional methods for predicting the performance of concrete and designing the composition of UHPC cannot be met.UHPC is a high-strength,high-workability concrete,and its performance prediction and composition design face greater challenges.Therefore,this paper uses an integration learning-PSO algorithm to predict the performance of UHPC and optimize its composition design.Firstly,this paper constructs a data learning set including the strength,slump,and composition of UHPC through extensive literature review and data analysis.Then,it uses existing integration learning models to analyze the strength and workability of UHPC.The study finds that the XGBoost integration learning model provides the best prediction performance for UHPC strength and workability.Next,this paper considers the effects of major factors such as the diameter and length of steel fiber on the integration learning model,and uses data enhancement to correct and improve the prediction accuracy.Furthermore,this paper explains the model using the SHAP method based on the analysis of the impact rules of the components of UHPC on material performance.It further analyzes the impact degree of different component contents on the strength and workability of UHPC.Finally,this paper uses the XGBoost-PSO model and particle swarm optimization algorithm to optimize the composition design of UHPC,using the comprehensive cost of UHPC cost and carbon emissions rights as control parameters.These methods provide a new idea for the composition design of UHPC.Overall,this paper successfully predicts the performance of UHPC using an integration learning-PSO algorithm and optimizes its composition design.These methods provide new ideas and technical support for the application of UHPC,and have important significance for improving the performance of concrete.