Predicting stress-dependent gas permeability of cement mortar with different relative moisture contents based on hybrid ensemble artificial intelligence algorithms. (19th September 2022)

• Record Type: Journal Article
• Authors: Chao, Zhiming; Wang, Mingyang; Sun, Yinuo; Xu, Xu; Yue, Wenhan; Yang, Chuanxin; Hu, Tianxiang
• Is Part Of: Construction & building materials. Volume 348(2022)
• Publication Date: 2022-09-19
• Subjects: Cement mortar -- Permeability -- Relative moisture content -- Machine learning algorithm -- Optimisation algorithm; Building materials -- Periodicals; 624.18
• DOI: 10.1016/j.conbuildmat.2022.128660
• Languages: English
• ISSNs: 0950-0618

Abstract

Highlights: 1890 laboratory gas permeability tests were conducted on cement mortar with different moisture saturations. Predictive models for the gas permeability were established based on hybrid ensemble artificial intelligence algorithms. The forecasting performance of the established machine learning models was compared and verified. Abstract: The gas permeability of cement mortar is a vital parameter for the reasonable design of relevant buildings. However, due to the large number of influence factors and the complex interaction mechanism, the correct prediction of the gas permeability for cement mortar is always a huge challenge. In this paper, a novel machine learning model was established by combining Mind Evolutionary Algorithm (MEA) and the ensemble algorithm of Adaptive Boosting Algorithm (ADA)-Back Propagation Artificial Neural Network (BPANN) to predict the stress-dependent gas permeability of cement mortar with different moisture contents based on the results of 1890 laboratory gas permeability experiments. The novel machine learning model combined by MEA and ADA-BPANN is firstly adopted in the prediction of the gas permeability of cement mortar. By comparing with the conventional machine learning algorithms, including Particle Swarm Optimisation Algorithm (PSO) and Genetic Algorithm (GA) tuned ADA-BPANN, MEA tuned Extreme Learning Machine (ELM) and Random Forest (RF), the superior performance of MEA tuned ADA-BPANN has been validated. This new algorithm is with higher predicting precision, shorter training time, and the avoidance of local optimum and overfitting. Secondly, sensitivity analysis was carried out by adopting this proposed novel model, which indicates that the impact of relative moisture content on the gas permeability is the highest, followed by confining pressure, cyclic time and confining pressure loading/unloading stage. Thirdly, an analytical equation was proposed to assess the gas permeability that allows the usage of machine learning skills for the practitioners with limited machine learning knowledge. The present research highlights the potential of the MEA tuned ADA-BPANN model as a useful tool to assist in preciously estimating the stress-dependent gas permeability of cement mortar with different moisture contents. This can provide huge help for the reasonable design of relevant engineering applications.