Improvement of prediction model of concrete crushing based on random forest algorithm

The use of efficient and accurate methods for crushing cement concrete will be a development trend in road maintenance and renovation. Based on orthogonal design methods, concrete pressure tests and hydraulic concrete crushing tests were conducted, and the influence of multi-dimensional parameters on the depth and diameter of concrete crushing was studied. At the same time, hydraulic parameters were optimized from an economic perspective. Finally, the random forest algorithm model was used to predict the crushing depth and diameter of concrete under hydraulic action. The prediction model is improved based on K-fold cross validation and grid search methods, and the three indices of model fit coefficient R2, root mean square error (RMSE), and mean absolute error (MAE) are used to comprehensively evaluate the prediction effect of the model. The results show that the effects of pump pressure and crushing time on the crushing depth and diameter of concrete are positive; concrete with high strength is more resistant to hydraulic crushing, and an optimal value of target distance exists for the crushing depth and diameter of concrete. The random forest algorithm model has high accuracy for predicting the crushing depth and diameter of concrete, and the method of K-fold cross-validation and grid search is used to improve the crushing accuracy of concrete. The study provides the theoretical basis and technical support for the promotion of the use of hydraulic crushing technology in road maintenance and demolition engineering.