As a cementitious composite, concrete’s property depends on the matrix generated from cement hydration and the dispersed phases such as aggregates. Compression strength is an important mechanics performance index of concrete quality, especially the High-performance Concrete (HPC). However, owing to the expensive cost of test and the existence of high-dimensional nonlinear mapping between compression strength and basic materials, it is uneasiness to precisely forecast the compression strength value of HPC by general formula method. In this research, a novel machine learning system, Genetic Algorithm and BP Neural Network (GA-BPNN) coupling algorithm, is offered to predict the compression strength of HPC. GA-BPNN coupling algorithm model used 181 groups of HPC mixture data to determine 8 factors affecting its compression strength (i.e., Water, Portland Cement, Waterbinder Ratio, Fine Aggregate Ratio, Air-entraining Agent, Fly Ash, Silica Fume, and Superplasticizer) as the input variables of the model, while compression strength was set as the output variable. In addition, 166 sets of training set data were segmented into training, validation and test set again, and BP neural network (BPNN) was compared with GA-BPNN to verify the generalization capacity of the model in this research. By forecasting the compression strength of 15 test sets, the average relative error is only 0.902%. Finally, the sensitivity of input variables of GA-BPNN model was analyzed by using Gray Relational analysis (GRA) method. Six models were established to research the impact of sensitivity and quantity of input variables on model performance by ignoring individual input variable. The research is shown that GA-BPNN model not only has the powerful nonlinear mapping ability of BPNN, but also has the global search optimization ability of GA, and showed stronger robustness and prediction potential in the assessment of compression strength value of HPC. The sensitivity analysis shows that, to compression strength of HPC, Cement, Water and Water-binder ratio has a sensitivity score of 0.8166, 0.70122, 0.66772, respectively while Fly Ash has the lowest sensitivity.

There is a complex high-dimensional nonlinear mapping relationship between the compressive strength of High-Performance Concrete (HPC) and its components, which has great influence on the accurate prediction of compressive strength. In this paper, an efficient robust software calculation strategy combining BP Neural Network (BPNN), Support Vector Machine (SVM) and Genetic Algorithm (GA) is proposed for the prediction of compressive strength of HPC. 8 features were extracted from the previous literature, and a compressive strength database containing 454 sets of data was constructed. The model was trained and tested, and the performance of 4 Machine Learning (ML) models, namely BPNN, SVM, GA-BPNN and GA-SVM, was compared. The results show that the coupled model is superior to the single model. Moreover, because GA-SVM has better generalization ability and theoretical basis, its convergence speed and prediction accuracy are better than GA-BPNN. Then Grey Relational Analysis (GRA) and Shapley analysis were used to verify the interpretability of the GA-SVM model, which showed that the water-binder ratio had the most significant influence on the compressive strength. Finally, the combination of multiple input variables to evaluate the compressive strength supplemented this research, and again verified the significant influence of water-binder ratio, providing reference value for subsequent research.