Optimized line balancing application considering demand forecast and artificial neural networks

Having a line balancing model linked to fluctuations in demand can directly contribute to the important reduction of costs linked to manufacturing. With organizations undergoing more and more transformations and facing high competitiveness, it is essential to adopt quantitative methods and optimized...

Descripción completa

Detalles Bibliográficos
Autores: Moraes, Nathalia Tessari, Corso, Leandro Luís
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2023
País:Brasil
Institución:Associação Brasileira de Engenharia de Produção (ABEPRO)
Repositorio:Revista Produção Online
Idioma:portugués
OAI Identifier:oai:ojs.www.producaoonline.org.br:article/4734
Acceso en línea:https://www.producaoonline.org.br/rpo/article/view/4734
Access Level:acceso abierto
Palabra clave:Line balancing
Demand forecast
Artificial intelligence
Balanceamento de linha
Previsão de demanda
Inteligência artificial
Descripción
Sumario:Having a line balancing model linked to fluctuations in demand can directly contribute to the important reduction of costs linked to manufacturing. With organizations undergoing more and more transformations and facing high competitiveness, it is essential to adopt quantitative methods and optimized processes that guarantee greater efficiency in resource management. Predicting market behavior is not a simple task, especially when there is high variability in demand. Thus, it is important to consider robust mathematical models with optimized configurations so that they are able to recognize patterns, in order to predict the sales volume with the least possible error. In view of this, the historical sales data of the five main products of a multinational company, those that represent the highest profit, were used, and the demands were calculated using the moving average, exponential smoothing, Box-Jenkins and RNA models. Afterwards, in order to choose the most accurate method, that is, the one with the lowest error, the errors of each of the methods used (RMSE, MAE and MAPE) were calculated in isolation, thus enabling the comparison between them. In view of the result obtained, the balancing of the assembly lines in which the products are produced was modeled, the number of employees for the expected demand was calculated, using a mathematical model of non-linear programming, having as main objective improve current efficiency by organizing lines. Thus, it was found that the ANNs represented greater accuracy for forecasting demand and that, in comparison with the current method used by the company, it proved to be a more reliable method of predicting quantities for resource planning. Afterwards, it was visualized that the balancing method developed brought important adjustments of resources, thus leading to an increase in the efficiency of the production lines by a percentage of 29%, which is considered quite satisfactory.