Aplicação de redes neurais para a previsão de séries temporais, referentes ao consumo de gás nitrogênio em uma empresa de margarina do estado do ceará

Abstract

In a world that is living in its fourth industrial revolution, or industry 4.0, the use of digitization processes have become fundamental to the survival of any company, regardless of its sector of operation, in view of the fact that these operations allow not only a greater knowledge of the company, with its own procedures and variables, but provide a commercial advantage over its competitors. Faced with this new type of challenge, companies have been looking for ways to follow up and forecast their main process variables, among several other approaches. Therefore, the work in question has a practical approach about a viable application for the prediction of the consumed value of nitrogen gas (N2), taking into account all the requisite processes of this type of material, in a margarine producing company, located in the state of Ceará. To carry out this activity, tools called neural networks were used, as well as the analysis and processing of real data of the company’s operation, which were collected digitally through the aid of specific hardware, which provided values in time, conditions of use of N2. It can be seen that the models applied acted satisfactorily with the proposed, obtaining future values of behavioral aspects for the conditions of use of the product in studies, allowing the company holding the data, a greater control in the application of this product.