CHARACTERIZATION OF PORTLAND CEMENT COMPOUNDS WITH HYPERSPECTRAL IMAGING

Date

2023-11-22

Author

Shakouri, Sahra

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

99
views

117
downloads

Cite This

Developing new and reliable methods for materials characterization is a progressing field of science and engineering. Hyperspectral imaging system is noted as one of the modern and proceeding techniques that can be employed in numerous subjects from remote sensing of earth and medical diagnoses to analyzing nanomaterials. This method is based on generating a spectral map for each spatial pixel of an image taken in any scale. Advantages like accuracy, rapidity, cost efficiency are some of the features that put this method one step ahead of the other possible options. All those advantages make hyperspectral imaging very suitable for cement analysis. As one of the most widely utilized building materials, Portland cement has a complex structure with distinct compounds. Each of its compounds has a unique set of reaction kinetics during the hydration process. Investigating each reaction requires a series of specialized and complicated tests, which makes finding more convenient ways of observing these details highly demanded. This is one of the first studies on the use of hyperspectral imaging in detecting cement compounds. The research utilizes reflectance hyperspectral spectroscopy in the visible and near-infrared spectral range (400 to 1000 nm) to evaluate the characteristics of cement compounds. Each compound is produced in the laboratory with high purity, and their physical, chemical, and mechanical properties are thoroughly examined alongside their hyperspectral signatures. After synthesizing these compounds, hyperspectral data are collected, forming the foundation for classifying randomly chosen cement clinkers. The accuracy of this classification was evaluated using the Deep Learning Toolbox in Matlab, specifically employing a Kernel Logistic Regression model. Notably, this model achieves an accuracy of 98.25% in classifying different types of cement through hyperspectral imaging.

Subject Keywords

Material Characterization, Hyperspectral Signature, Hyperspectral Imaging Analysis, Cement Synthesis

URI

https://hdl.handle.net/11511/107843

Collections

Graduate School of Natural and Applied Sciences, Thesis