Prediction of the antifreeze of the concrete structure based on random forest and wavelet neural network

[摘要] The durability of concrete is the focus of research in the field of engineering, and the resistance of concrete is one of the important indicators of concrete durability. Based on the prediction of early resistance to concrete based on the random forest combined with the wavelet neural network algorithm, 12 factors affecting concrete antifreeze were selected from the ratio level of concrete material, and the relative dynamic elastic mode was used as the evaluation index of concrete resistance, and the importance evaluation and feature variable selection of influencing factors were used by random forest, and a concrete antifreeze prediction model based on RF-WNN (random forest-wave neural network) was established. The example analysis is carried out by taking the project project as an example, and the calculation of the wavelet neural network and BP neural network prediction model without the influence factor screening is compared to obtain the mean square root error and the fit superiority respectively. The results show that the prediction results of RF-WNN model are closer to the actual value and the prediction accuracy is higher, and the proposed RF-WNN prediction model provides an effective method for achieving concrete antifreeze prediction.

[学科分类] 材料科学（综合）