PIPE: Um preditor de tempos de viagem usando fluxo contínuo de trajetórias

• Ano de defesa: 2018
• Autor(a) principal: Nascimento, Samara Martins
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Preditor; Trajetórias; Fluxos contínuos de trajetórias
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/38808

Traffic patterns or traffic anomalies can be understood from analyzes related to moving objects. These analyzes can be performed both in historical context, as in real time, allowing you to see about traffic, capturing or detecting its changes, dynamically, events or incidents as they happen. Within this context, trajectory data are of fundamental importance in the characterization of the behavior of moving objects. However, computing models that can predict the travel time of objects, almost in real time, is considered a large challenge. The main impediment is related to the need to present traffic-related changes when new continuous flows of trajectories are received. In addition, despite its great applicability, the use of trajectory data is not excluded from problems, which justifies its extensive exploration in the current literature, presenting studies on the processing of large volumes of data, handling of errors and inaccuracies and the construction of predictive models that can, for example, estimate the total time to reach a destination from a given origin. This work tries to face the challenge of constructing a new model of prediction, which is able to compute results about the travel time of moving objects, when they are reported continuous flows of trajectories. Within this context, this research proposes the prediction model called PIPE: A Predictor of Travel Times using Continuous Trajectory Flow, which can be used to estimate the travel time of a moving object to travel through a specific street segment given one hour of the day. Thus, this thesis seeks to answer, in general, two most important research questions: (i) Is it possible to create a prediction model, to estimate the travel time of the objects, considering a set of trajectories?; and also, (ii) How to construct a model that performs incremental maintenance, given the receipt of continuous flows of trajectories, and generate, as a result, prediction functions in real time?. The PIPE model is responsible for generating a prediction function and updating it, given the dynamic reception of the data. The experimental evaluation of this work is conducted for two sets of data real. The experimental results, which validated the proposed solution, show analyzes related to the processing time to construct and update the derivable function, and also discuss the results related to the accuracy of the solution.