Loading...
Publication
Exploração de modelos preditivos para sistemas de controlo de acesso
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.97 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
A Agenda NEXUS, sob a liderança da Administração dos Portos de Sines e do Algarve (APS),
é um projeto destinado a impulsionar a transformação digital e sustentável no setor dos
transportes e logística. Esta mudança abordará áreas como a aplicação da Inteligência
Artificial nas atividades portuárias e nos processos de transporte e logística.
O controlo de acessos será uma das áreas abordadas e desempenha um papel fundamental
na segurança de dados e instalações, permitindo que apenas pessoas autorizadas tenham
acesso a recursos críticos. O controlo de acessos desempenha um papel fundamental na
manutenção da segurança, integridade e eficiência das operações portuárias.
Os modelos preditivos podem desempenhar um papel crucial no controlo de acessos nas
áreas portuárias, melhorando a eficiência e a segurança das operações. Estes modelos
podem ajudar em diversas áreas, como por exemplo: através da identificação de padrões
anómalos, analisando dados históricos de acesso, como horários, locais e tipos de acesso.
Podem também efetuar previsões de Fluxos de Tráfego, com base em dados históricos e em
informações em tempo real. Isso ajuda na gestão do fluxo de veículos e pessoas, evitando
congestionamentos e melhorando a eficiência das operações.
Este trabalho tem como principais objetivos o estudo a avaliação de diversos modelos
preditivos para um conjunto de dados fornecido pela APS, por forma a se conseguir prever o
número de pedidos de acessos às instalações e conseguir ajustar o número de servidores
necessários para garantir o correto funcionamento do sistema.
Começou-se por efetuar uma revisão da literatura, investigando conceitos como controlo de
acessos, modelos preditivos, machine learning e casos de estudo onde já tinham sido
efetuados projetos similares. Como guia para a aplicação dos algoritmos de machine learning
seguiu-se a metodologia CRISP. Por fim, na aplicação dos algoritmos utilizou-se o Jupyter
Notebook para programar em Python.
Após configuração, aplicação e avaliação de diversos algoritmos, conseguiu-se chegar a uma
versão final de algoritmo que permite efetuar previsões do número de acessos por hora, o que
permitirá à entidade gestora dos servidores realizar um planeamento da alocação de
servidores, proporcionando a oportunidade de reduzir custos.
The NEXUS Agenda, under the leadership of the Sines and Algarve Port Administration (APS), is a project aimed at driving digital and sustainable transformation in the transport and logistics sector. This change will address areas such as the application of Artificial Intelligence in port activities and transportation and logistics processes. Access control will be one of the areas addressed and plays a fundamental role in the security of data and facilities, allowing only authorized individuals to access critical resources. Access control plays a crucial role in maintaining the security, integrity, and efficiency of port operations. Predictive models can play a crucial role in access control within port areas, enhancing both the efficiency and security of operations. These models can assist in various areas, such as identifying anomalous patterns by analyzing historical access data, including times, locations, and types of access. They can also make Traffic Flow Predictions, using historical data and real-time information. This helps manage the flow of vehicles and people, preventing congestion and improving operational efficiency. The main objectives of this work are to study and evaluate various predictive models for a dataset provided by APS, to predict the number of access requests to the facilities and to adjust the number of servers needed to ensure the proper functioning of the system. The work began with a literature review, exploring concepts such as access control, predictive models, machine learning, and case studies where similar projects had already been implemented. The CRISP methodology was followed as a guide for applying machine learning algorithms. Finally, in the implementation of the algorithms, Jupyter Notebook was used for programming in Python. After configuring, applying, and evaluating various algorithms, a final algorithm version was developed that enables hourly access forecasts. This will allow the server management entity to plan server allocation, providing an opportunity to reduce costs.
The NEXUS Agenda, under the leadership of the Sines and Algarve Port Administration (APS), is a project aimed at driving digital and sustainable transformation in the transport and logistics sector. This change will address areas such as the application of Artificial Intelligence in port activities and transportation and logistics processes. Access control will be one of the areas addressed and plays a fundamental role in the security of data and facilities, allowing only authorized individuals to access critical resources. Access control plays a crucial role in maintaining the security, integrity, and efficiency of port operations. Predictive models can play a crucial role in access control within port areas, enhancing both the efficiency and security of operations. These models can assist in various areas, such as identifying anomalous patterns by analyzing historical access data, including times, locations, and types of access. They can also make Traffic Flow Predictions, using historical data and real-time information. This helps manage the flow of vehicles and people, preventing congestion and improving operational efficiency. The main objectives of this work are to study and evaluate various predictive models for a dataset provided by APS, to predict the number of access requests to the facilities and to adjust the number of servers needed to ensure the proper functioning of the system. The work began with a literature review, exploring concepts such as access control, predictive models, machine learning, and case studies where similar projects had already been implemented. The CRISP methodology was followed as a guide for applying machine learning algorithms. Finally, in the implementation of the algorithms, Jupyter Notebook was used for programming in Python. After configuring, applying, and evaluating various algorithms, a final algorithm version was developed that enables hourly access forecasts. This will allow the server management entity to plan server allocation, providing an opportunity to reduce costs.
Description
Keywords
Agenda Nexus Controlo de Acessos Atividades Portuárias Modelos Preditivos