Diseño y simulación de un sistema de control de calidad basado en redes neuronales para la clasificación paneles decorativos. caso empresa del sector maderero.

Abstract

The large amount of decorative laminated panels that are manufactured in companies dedicated to the production of wood panels, in conjunction with meeting the expectations of customers, has generated the obligation to maintain quality control during their production in order to ensure a reliable product. In the final stage of the board lamination process, a quality check is carried out, sometimes very common errors are not detected, due to the speed of the production line and the visual verification is prone to errors. Misclassification generates complaints from both the domestic and export markets, causing reprocessing and unanticipated long delays to customers. The first part of the research establishes the research question, the objectives and the hypothesis, followed by a theoretical referential where the conceptual foundations linked to the study are established. Subsequently, it is shown how the quality control system was designed and simulated by means of a neural network, to identify the class B boards, which present quality failures either of torn paper or glued paper, and to compare the benefits that it would add to the production process, with the boards identified and not identified during its production, whose information has been collected from 2021 to 2022 by means of production records. To conclude the research, a proposal is presented, consisting of a budget for a possible implementation of the board classification system.