An efficient model for predicting setting time of cement based on broad learning system

• Published in: Applied soft computing 2020-11, Vol.96, p.106698, Article 106698
• Main Authors: Guo, Jifeng, Wang, Lin, Fan, Kaipeng, Yang, Bo
• Format: Article
• Language: English
• Subjects: Broad learning system; Clinker composition; Final setting time; Initial setting time; Physical properties
• ISSN: 1568-4946; 1872-9681
• DOI: 10.1016/j.asoc.2020.106698

Cement is the main building material in the construction industry. Its setting time directly affects the setting time and strength of concrete, which further affects construction schedule and building quality. However, traditional measurement technology not only has high labor intensity and high time consumption, but has a high technical requirement. Various human factors, such as insufficient operation, will result in great errors in the measurements. The accurate prediction of setting time enables manpower savings, avoids large errors caused by insufficient operation, and guides the production of high-performance cement. In this paper, an efficient model based on the broad learning system is proposed to predict the initial and final setting time. It is committed to directly predicting setting time from clinker composition and physical properties, which is of great significance to the optimization of clinker formula. The experimental results show that it can accurately predict the setting time and behave good generalization ability, which addresses the problem of labor intensity in measurement and saves many resources. In addition, the broad learning system can rapidly build a setting time prediction model with few errors, satisfying industrial demands for the rapid modeling of various specialty cements. •The model is the first to predict the setting time in terms of cement properties.•This study is the first to predict the setting time of cement using BLS.•It solves the arduous issue in measurement, achieving rapid modeling for cement.