Use of Artificial Neural Network for Analyzing the Contributions of Some Kinematic Parameters in the Polishing Process of Porcelain Tiles

Abstract

The glossiness of porcelain tiles is a highly appreciated feature by its consumers and so it became a crucial aspect on the quality control of this product. The polishing process, which enhance the porcelain tile with the glossiness effect, is a complex process regarding the individual gloss values of each region of the tile and the glossiness final pattern. Due to this complexity, the use of artificial neural networks is a promising tool to build a regression model of glossiness prediction. This paper presents the performance of eight neural networks designed to predict glossiness in a specific point of the tile’s surface. Each model has different inputs and so the aim of this work is to quantify the influence of the inputs on the performance metric of the neural network. The identified input parameters which had the highest influence on the gloss value were the number of abrasive contacts, average scratching speed and standard deviation of scratching direction. This study also intends to provide scientific basis to future works on employing neural networks to predict glossiness of porcelain tile after the polishing process.