Discovery of Transport Operations from Geolocation Data

• Autor(a) principal: Jorge Alberto da Mota Vieira Tavares
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: https://hdl.handle.net/10216/137325
• topic: Engenharia electrotécnica, electrónica e informática; Electrical engineering, Electronic engineering, Information engineering

Geolocation data identifies the geographic location of people or objects, and is fundamental for businesses relying on vehicles such as logistics and transportation. With the advance of technology, collecting geolocation data has become increasingly accessible and affordable, raising new opportunities for business intelligence. This type of data has been used mainly for characterizing the vehicle in terms of positioning and navigation, but it can also showcase its performance regarding the executed activities and operations. The proposed approach consists on a multi-step methodology that receives geolocation data as an input and allows the analysis of the business process in the end. Firstly, the preparation of the data is applied to handle a number of issues related to outliers, data noise, and missing or erroneous information. Then, the identification of stationary events is performed based on the motionless states of the vehicles. Next, the inference of operations based on a spatial analysis is performed, which allows the discovery of the locations where stationary events occur frequently. Finally, the identified operations are classified based on their characteristics, and the sequence of events can be structured into an event log. The application of process mining techniques is then possible and the consequently extraction of process knowledge. The steps of the methodology can also be used separately to tackle specific challenges, giving more flexibility to its application. Three distinct case studies are presented to demonstrate the effectiveness and transversality of the solution. Real-time geolocation data streams of buses from two distinct public transport networks are used to demonstrate the detection of vehicle-based operations and compare the distinct approaches proposed by this work. The buses operations produce a structured sequence of events that describes the behaviour of the buses. This behaviour is mapped through the application of process mining techniques uncovering analysis opportunities and discovering bottlenecks in the process. Geolocation data from an international logistics company is exploited for monitoring logistics processes, namely for detecting vehicle-based operations in real time, showing the effectiveness of the proposed solution to solve specific industry problems. The results of this work reveal new possibilities for geolocation data and its potential to generate process knowledge. The exploitation of geolocation data in the public transport and logistics contexts poses as an opportunity for improving the monitoring and management of vehicle-based operations. This can lead to into improvements in the process efficiency and consequently higher profit and better service quality.