ABSTRACT
Data analytics is one of the most important concepts that drives industry 4.0. Most industries give a limited use to the process data. Specifically, foundry industries that involve complex processes that can be enhanced by data analytics. In this context, this paper explores data analytics methods for foundry industries with sand casting to minimise the percentage of product defects by controlling sand parameters, such as compactability and compressive strength. The present work contributes with two models to predict: (i) sand parameters; and (ii) casting rejections. First, we apply a Multiple Linear Regression to predict the sand parameters using the initial moisture sand, the amount of water introduced, and the bentonite. Then, a rule-based model is applied to predict the percentage of casting defects based on sand parameters data. Experiments with an industrial dataset show that the proposed methods improve the accuracy of the predictions when compared to a baseline approach.
