Prediction Explosive Spalling Of Ultra-High Performance Concrete Under Fire Based On Artificial Neural Network

Ultra-high performance concrete is superior to ordinary concrete in many aspects such as strength and durability,but its microstructure is more dense and its high temperature anti-explosion performance is relatively poor,and building fire will cause immeasurable casualties and economic losses.Therefore,it is of great significance to predict the high-temperature explosive spalling of ultrahigh performance concrete,but the mechanism of high-temperature explosive spalling is still not clear and cannot be predicted by traditional simulation methods.Therefore,this paper adopts a popular machine learning method-artificial neural network(ANN),mainly to predict ordinary concrete,high performance concrete and ultra-high performance concrete,and further studies the influence of hybrid fiber on the high temperature resistance of concrete.The main research contents are as follows:1.This paper introduces two kinds of artificial neural network(ANN)models,one is based on concrete mix ratio model(ANN1),the other is based on compressive strength model(ANN2),to evaluate the anti-explosion spalling ability of concrete.648 and 634 test data collected in the literature were used to train the two neural network models proposed in this paper.2.High performance concrete(HPC),ultra-high-performance concrete(UHPC)and plain concrete(PC)mixed with polypropylene fiber and steel fiber were designed,and 96 groups of fire tests were carried out.By comparing and analyzing the test results with the model prediction results,the prediction accuracy of ANN1 and ANN2 reached 100% and 98%,respective.Compared with the method proposed by Mugume and Horiguchi to predict concrete spalling by explosion,the ANN prediction model has higher applicability and accuracy.3.Combined with the artificial neural network model input data set and the high temperature test results,the influences of single polypropylene fiber,hybrid polypropylene fiber and steel fiber,compressive strength,moisture content and specimen size on the high temperature anti-explosion performance of concrete are analyzed.