Descriptive and predictive analytics of the cement and concrete production process
Fecha
2022-06
Autores
Madrid Peláez, Juan Pablo
Cardeño Luján, Salomón
Título de la revista
ISSN de la revista
Título del volumen
Editor
Resumen
The paper uses both a Machine Learning and a combined Taylor Modeling with Heuristics
approach to predict the Compressive Strength, the main characteristic to determine the
physicochemical behaviors of concrete and cement. Implementing a data-based approach, can
result in a reduction of the variability of the process by improving the overall reliability. The
AdaBoost, ANN and SVM algorithms were implemented, where the AdaBoost performed
the best in terms of fitness, error, and predictions, which was expected due to its k-folds
nature. To counter the black-box nature of Machine Learning, a Taylor Modeling algorithm
was implemented to build the mathematical model of the data with the use of an evolutionary
algorithm to find the best parameters for the model.