|
Title |
Water-Cement Ratio Estimation of Cementitious Materials Using Electrochemical Impedance Spectroscopy and Machine Learning
|
|
Authors |
박주혜(Joohye Park) ; 김도윤(Doyun Kim) ; 심소현(Sohyun Shim) ; 홍진영(Jinyoung Hong) ; 최하진(Hajin Choi) ; 진승섭(Seung-Seop Jin) |
|
DOI |
https://doi.org/10.4334/JKCI.2021.33.4.353 |
|
Keywords |
전기화학적 임피던스 분광법; 기계학습; 품질관리; 굳지 않은 콘크리트 electrochemical impedance spectroscopy; machine learning; quality control; fresh concrete |
|
Abstract |
The objective of this research was to develop a new concept of material characterization technology in cementitious materials, using machine learning algorithms and electrochemical impedance spectroscopy (EIS). Although several studies on cementitious materials have used EIS, the accuracy of equivalent circuit models that imply a relation between the circuitry parameters (resistance and capacitance) and micro-structure of cementitious materials is limited. To address this, the parameters were theoretically normalized and machine learning algorithms were applied. The EIS tests were conducted using 140 cement paste samples, and four different machine learning algorithms were evaluated to predict the relation between the parameters and w ater-to-cement ratio. The results demonstrated that Gaussian process regression using an ARD exponential kernel possibly predicts water-to-cement ratio within the minimal error (MAE 5.44; RMSE 1.82). Therefore, the proposed method has great potential in the evaluation of the material characterization of cementitious materials.
|