Premium
Neural network model for predicting early strength of cementing materials
Author(s) -
Wang Jian,
Qian Chunxiang,
Zhang Huimin,
Qu Jun,
Guo Jingqiang
Publication year - 2018
Publication title -
structural concrete
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.912
H-Index - 34
eISSN - 1751-7648
pISSN - 1464-4177
DOI - 10.1002/suco.201700179
Subject(s) - mortar , artificial neural network , shrinkage , approximation error , curing (chemistry) , cementitious , materials science , fly ash , void ratio , correlation , composite material , cement , mathematics , statistics , computer science , artificial intelligence , geometry
After judging whether the early strength of mortar after steam curing can meet the requirements of the prefabricated products, the combination of the cementitious materials can be judged. The main factors influencing early strength of mortar are analyzed by gray correlation analysis and neural network model is established through MATLAB, and then the early strength value and relative error of steam curing mortar can be predicted. From the analysis of gray correlation, the correlation factors of the volume ratio between CSH and unhydrated particle is 0.58, the correlation factors of bulk density and early strength is 0.62. The neural network model is established as the input variable, the average relative error of the test sample is 8.63%, and the average relative error of the sample to be tested is 6.72%. The prediction results show that the suitable dosage of fly ash is 6%, the suitable dosage of slag powder is 10%, and the amount of limestone powder should be controlled within 10%. The combination of gray correlation analysis and neural network model is feasible.