Monitor concrete moisture level using percussion and machine learning

The durability of underwater and hydraulic concrete structures is highly dependent on their moisture content, which makes the evaluation of moisture contents of great significance in ensuring the proper functioning of these structures. This paper develops a novel percussion-based method to identify the moisture level of concrete. The method of percussion refers to tapping and listening. As a popular acoustic feature used in the field of speech recognition, the Mel-Frequency Cepstral Coefficients (MFCCs) are used in this paper as the features of impact-induced sound. In addition, a microphone was employed to obtain the impact-induced sound signals and a support vector machine (SVM) based machine learning were utilized to classify the different moisture content in concrete. The experimental results demonstrate that the proposed percussion method can identify different moisture levels in concrete with accuracy more than 98%. In comparison to traditional methods for evaluation of moisture content, the proposed percussion method is easy to operate and requires no sensor installation. (C) 2019 Elsevier Ltd. All rights reserved.