Análise e simulação baseada em agentes de rotas aéreas

• Ano de defesa: 2017
• Autor(a) principal: Sandro Jerônimo de Almeida
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Consumo de combustível; Desempenho de aeronaves; Rotas aéreas; Simulação multiagente; Bigdata; Dados ADS-B; Mineração de dados; Metodos de simulação; Aerovias; Aeronaves Consumo de combustível; Mineração de dados (Computação); Engenharia mecânica
• Link de acesso: http://hdl.handle.net/1843/BUOS-AU6EVQ

Airlines and Governments have worked to optimize the use of their resources and operate efficiently, given the growing flight demand, increasing oil price, operational costs, and limited infrastructures. Some works have analyzed the performance of a limited set of flights, while others have explored specific problems, such as the flight trajectory optimization and the flight schedule problems. The objective of this work is the study and the development of technologies that can lead airlines to improve the efficiency of their flights operations. By using data analysis, mining and visualizations techniques, we conducted a performance study of 36,190 worldwide flights. This analysis showed relations among the flights data, for example, the aircrafts distance flown per flight phases and airports. These identified relations and extracted performances metrics can serve as reference for airlines consider in their operations improvement studies. A contribution of this work is the systematization of a process and implementation of a scalable computational tools for data: obtaining, extraction, transformation, cleaning, enrichment, inconsistency treatment, processing, data-mining, visualization of large flights datasets and operational flight data patterns discovery. A multiagent system was also proposed and developed aiming: i) to reproduce historical flights data through their trajectories and the computation/estimation of operational metrics, such as flight distance flown and fuel burned. ii) to simulate hypothetical scenarios and operational strategies. Case studies showed that airlines could save about 15% of fuel and increase seat availability in an unconstrained and best-case scenario. The obtained results reinforce two hypotheses: i) the study showed the possibility of using multiagent system technology for scalable and distributed computation of large datasets. In this case, the multiagent system can be seen as an alternative tool to Big Data processing. ii) the obtained results by using the simulator reinforce the idea that a holistic approach which combine different techniques, tend to be more efficient than a punctual technique application.