| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 3.93 MB | Adobe PDF |
Orientador(es)
Resumo(s)
Durante 2020 e 2021, vários setores foram fortemente impactados pela pandemia,
incluindo a indústria aeronáutica. Surge então a necessidade de reforçar toda a estrutura
de empresas que contribuem para o bom funcionamento dos aeroportos. Nomeadamente
a otimização de processos da cadeia de abastecimento nas empresas fornecedoras dos
aeroportos. A gestão de stocks do fornecimento de catering do aeroporto de Lisboa não é
exceção. Atualmente são reportados elevados valores de rotura de stock, devido à gestão
manual existente.
Este trabalho tem como objetivo modelar o comportamento da procura através do
número de passageiros do aeroporto, para construir políticas de gestão de armazéns. Surge
assim a oportunidade de estudar o comportamento da procura de material e averiguar se
este é afetado pelo tráfego aéreo.
Para alcançar os objetivos, recorreu-se à análise de séries temporais e a métodos de
previsão através de modelos de regressão, uma aplicação de machine learning. Tendo
como ponto de partida as séries temporais, é encontrada uma relação entre duas variáveis
ao longo do tempo, o fluxo de passageiros no aeroporto e a procura de matéria prima.
São propostos vários modelos de regressão, incluindo o modelo de regressão linear por
troços, para prever a procura, tendo como variável explicativa o número de passageiros
do aeroporto de Lisboa. Através desta previsão, estudaram-se duas políticas de gestão
de stocks. Para tal, foram estimados os custos associados à cadeia de abastecimento. A
primeira política é um modelo determinístico, onde a procura é assumida como constante.
A segunda tem em conta a variabilidade da procura e a aplicação do algoritmo de Hadley-
Whitin.
Os resultados mostram como a procura de material está fortemente relacionada com
o número de passageiros do aeroporto de Lisboa. Esta relação é descrita no modelo de regressão
linear por troços. A construção de uma política de gestão de stocks que considera
a variabilidade da procura e admite um stock de segurança considerável descreve uma
solução adequada para a resolução do problema.
During 2020 and 2021, several sectors were heavily impacted by the pandemic, including the aviation industry. There is a need to strengthen the entire structure of companies which contribute to the proper functioning of airports. In particular, the optimization of supply chain processes at airport supplier companies. Lisbon airport catering supply stock management is no exception. Currently, high of stock-out values are reported, due to the existing manual management. This work aims to model demand behaviour through the number of passengers at Lisbon airport, in order to build warehouse management policies. This gives rise to the opportunity to study demand behaviour and find out if it is affected by air traffic. To achieve these objectives, time series analysis and prediction methods where used through regression models, an application of machine learning. Taking time series as a starting point, a relationship between the two variables over time is found. Several regression models are proposed, including the linear regression model by sections, to forecast demand, using the number of passengers at Lisbon airport as the explanatory variable. Through this forecast, two stock management policies were studied. To this end, the costs associated with supply chain were estimated. The first policy is a deterministic model, where demand is assumed to be constant. The second takes into account the variability of demand and the application of the Hadley-Within algorithm. The results show how demand is strongly related to the number of passengers at Lisbon airport. This relationship is described in the piece-wise linear regression model. A stock management policy which considers the variability of demand and admits a considerable safety stock describes an adequate solution to solve the problem.
During 2020 and 2021, several sectors were heavily impacted by the pandemic, including the aviation industry. There is a need to strengthen the entire structure of companies which contribute to the proper functioning of airports. In particular, the optimization of supply chain processes at airport supplier companies. Lisbon airport catering supply stock management is no exception. Currently, high of stock-out values are reported, due to the existing manual management. This work aims to model demand behaviour through the number of passengers at Lisbon airport, in order to build warehouse management policies. This gives rise to the opportunity to study demand behaviour and find out if it is affected by air traffic. To achieve these objectives, time series analysis and prediction methods where used through regression models, an application of machine learning. Taking time series as a starting point, a relationship between the two variables over time is found. Several regression models are proposed, including the linear regression model by sections, to forecast demand, using the number of passengers at Lisbon airport as the explanatory variable. Through this forecast, two stock management policies were studied. To this end, the costs associated with supply chain were estimated. The first policy is a deterministic model, where demand is assumed to be constant. The second takes into account the variability of demand and the application of the Hadley-Within algorithm. The results show how demand is strongly related to the number of passengers at Lisbon airport. This relationship is described in the piece-wise linear regression model. A stock management policy which considers the variability of demand and admits a considerable safety stock describes an adequate solution to solve the problem.
Descrição
Palavras-chave
Séries temporais Machine Learning Regressão linear por troços Gestão de stocks Algoritmo de Hadley-Whitin
