Concrete Performance Prediction And Mix Proportion Design Based On Machine Learning And Optimization Algorithm

With the development of society,modern buildings become more and more complex,and there are more and more requirements for the performance of concrete,such as strength,workability and Concrete crack resistance of mass concrete.The complexity of concrete components leads to the nonlinearity and discreteness of concrete performance.The traditional concrete performance prediction and mix ratio design methods based on the fitting of test results are increasingly unable to meet the needs of use.Machine learning has high robustness and data fitting ability,and ensemble learning is an excellent method in machine learning.In this thesis,the machine learning method is used to predict the performance of concrete,the idea of integrating learning finite element PSO(particle swarm optimization algorithm)and the structural design requirements are adopted to directly optimize the concrete mix proportion design,which can provide a new idea for the concrete mix proportion design.In this thesis,the data sets of concrete composition,strength and slump are constructed,and the integrated learning model is established and compared with the traditional learning model.It is found that the xgboost(extreme gradient lifting)integrated learning model has the best prediction performance.The model is explained by Shapley additional explanning method,and the influence of different factors on the workability of concrete such as compressive strength and fluidity is analyzed.Through the mix proportion and its strength parameters,the calculation parameters of the temperature field of mass concrete are determined according to the concrete performance test and empirical formula.The distribution of the temperature field and stress field of concrete structure under different mix proportion is obtained through the finite element calculation model.Different constraints are designed by using the concrete cost,combined with the concrete comprehensive cost of carbon emission right and the concrete temperature stress as the objective function,Xgboost PSO model is used to design the mix proportion of concrete materials.