The purpose of this paper is to present the result of a systematic literature review regarding the application and development of forecasting models in the industrial context, especially the context of manufacturing processes and operations management. The study was conducted considering the preparation of an established research protocol to know, discuss, and analyze the main approaches adopted by researchers in the field. To achieve this objective, we analyzed 354 recent papers published in periodicals between 2008 and 2018. This paper makes three main contributions to the field: (i) it presents an updated portfolio of prediction models in the industrial context, providing a reference point for researchers and industrial managers; (ii) it presents a characterization of the field of study through the identification of publication vehicles, frequency, and the principal authors and countries related to the development of research on the theme; (iii) it proposes a unified framework, listing the characteristics of the prediction models with their respective application contexts, identifying the current research directions to provide theoretical aids for the development of new approaches to forecasting in industry. The results of this study provide an empirical base for further discussions on studies that focus on forecasting in the industrial context.
O objetivo deste estudo foi ajustar um modelo de previsão combinado para prever o comportamento de um sistema de descarga de minério em um terminal portuário. Foram utilizadas observações de um mês de operação, coletadas a cada hora, totalizando 720 observações. Os procedimentos metodológicos foram realizados em três etapas: (i) ajuste dos modelos concorrentes por meio da metodologia de Box-Jenkins; (ii) combinação dos modelos selecionados pelos métodos de média aritmética, Least Squares Weights e Inverse Mean Squared; (iii) comparação das medidas de acurácia de cada previsão para seleção do melhor modelo. Foram obtidos 20 modelos concorrentes, dos quais foram selecionados três para a etapa de combinação. O melhor modelo foi obtido pelo método de combinação Inverse Mean Squared apresentando um erro absoluto médio de 2,017%. Os métodos de combinação proporcionaram um aumento da acurácia das previsões, gerando subsídios à tomada de decisão para melhorias no planejamento do processo produtivo.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.