Forecasting spare parts demand using an artificial neural network

• Autor(a) principal: Castro Soares, Laura Maria
• Data de Publicação: 2022
• Outros Autores: Gonçalves Castro Silva, Ana Cristina, Castro Silva, José de, Souza Santos, Pedro Vieira
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Sistemas & Gestão
• Texto Completo: https://www.revistasg.uff.br/sg/article/view/1806

Regarding the maintenance of assets, the demand forecasting for spare parts is an important condition in inventory management, aiming to reduce costs and avoid product obsolescence. The use of more accurate predictive methods is a fundamental tool against the lack of parts and over-stocking. Thus, the present work aims to evaluate the performance of an artificial neural network in predicting the demand for spare parts in a tractor maintenance department. Moreover, used the analysis of the absolute mean percentage error of the forecast as a method of evaluation and monitoring. Initially, a survey was made of the main theoretical aspects related to inventory management and demand forecasting methods. Subsequently, Elman networks were selected for the study. The selection of spare parts to be analyzed was done through inventory management tools to explore important items for the department. Using the proposed methodology, the results showed that the neural networks have a good application in this field, because, in addition to presenting configurations with acceptable errors, on several occasions the network hit peaks with higher and lower demands, an important analysis for the management of inventories.