REPOSITÓRIO P.PORTO
Scientific Repository of the Porto Polytechnic
Recent Submissions
Cinema e fotografia em residência artística: Vila do Conde
Publication . ; Ferreira, José Quinta; Leal, João; Cortesão, Maria João
Trabalhos resultantes da residência artística do Mestrado em Cinema e Fotografia
Work from the artistic residency of the Master in Film and Photography
Sistemas de pictogramas na ativação da identidade visual: o caso dos mercados de Matosinhos
Publication . Teixeira, Maria João Pinto; Fernandes, Marta
No âmbito da conclusão do Mestrado em Design, surge esta proposta de Projeto focada no Design de Pictogramas que visa melhorar a identificação dos Mercados de Matosinhos. Deste modo, será apresentada uma metodologia que pretende examinar diversos campos teóricos tais como: o Design e identidade de espaços, a Identidade Municipal, o Design de Pictogramas, e a Sinalética. O projeto sustenta-se, deste modo, em diversas metodologias como a analise literaria, estudos de caso, análise de campo, entrevista e metodologia projetual.
Na escolha do tema, foi fundamental estudar os mercados de Matosinhos, fazer uma recolha de informação relativamente a identidade visual atual, bem como uma visita e análise dos espaços.
Todos estes passos, vêem-se refletidos na concretização do projeto assente num Sistemas de Pictogramas para a ativação da identidade visual.
Coexistências transtemporais - Estudo fotográfico sobre a paisagem e a cartografia das linhas de defesa do Alto Minho
Publication . Oliveira, Hernâni Jorge Martins de; Baptista, Adriana; Alves, Cesário
O projeto Coexistências Transtemporais tem como objetivo refletir sobre a forma como o tempo se inscreve na paisagem, partindo através de processos investigativos e criativos da relação entre cartografia e imagem fotográfica. Tendo por base plantas militares elaboradas por Manuel Pinto de Vilalobos no início do século XVIII, foram revisitados os territórios por ele cartografados, com especial atenção às povoações fortificadas junto ao rio Minho.
Através da observação desses lugares no presente, procurou-se analisar como diferentes épocas coexistem no mesmo espaço, tanto nas formas construídas como nos vestígios que permanecem ou desaparecem. A paisagem (real ou construída) foi, aqui, entendida como um palimpsesto, onde camadas sucessivas de tempo se acumulam, se apagam parcialmente e se reescrevem. A fotografia foi utilizada como meio para observar, comparar e relacionar essas camadas temporais, revelando contrastes, permanências e transformações.
O trabalho combina investigação teórica com prática visual, propondo uma leitura do território como um lugar de cruzamento entre passado e presente e a fotografia como uma forma de elaborar memórias e chamar a atenção sobre o que não é percetível.
Multi-Agent Framework for Explainable and Context-Aware Forecasting with Reinforcement Learning and Dynamic Retraining in Smart Buildings
Publication . GOMES, LETÍCIA MILENE CONCEIÇÃO; Vale, Zita Maria Almeida do; Teixeira , Brígida Constança Correia
The accelerated growth of photovoltaic installations and evolving patterns of energy consumption have reshaped the global energy landscape. Solar energy systems continue to expand due to declining costs, improved efficiency, and strong policy commitments to sustainability. In parallel, global efforts to replace fossil fuels with renewable energy sources have led to exponential increases in solar energy capacity. This growth has fostered smart energy communities, with smart buildings acting as intermediaries between community-level coordination and individual end-users. However, tools to help users understand, optimize, and adapt their energy consumption remain limited. End-users often lack the technical expertise to interpret forecasts or act upon these
forecasts effectively. Most machine learning models used for prediction operate as black boxes, limiting user trust and actionable insights. Furthermore, the absence of mechanisms to justify or adapt predictions leaves systems vulnerable to error propagation and misaligned decision-making. The need for explainable, automated, and context-aware energy management frameworks is therefore critical, particularly those that combine datadriven
learning with interpretable rule-based reasoning. Multi-agent systems have emerged as a powerful tool for organizing and coordinating distributed tasks in complex environments. By delegating responsibilities to specialized agents capable of autonomous operation and communication, these systems enable decentralized processing, continuous adaptation, and modular expansion. In the energy domain, this solution is particularly well-suited for handling the heterogeneity of data sources, forecasting needs, and contextual behaviors. Its use ensures that various functionalities, such as data handling, decision-making, model updating, and result
dissemination, can be managed concurrently and cooperatively. This dissertation proposes a modular forecasting system to support energy management, specifically designed for smart buildings, which are a key component of smart energy communities. It is designed to deliver accurate and adaptive forecasts, provide explainable
insights tailored to diverse user knowledge profiles, and maintain sustained model quality under changing operational conditions, thereby supporting informed decision-making. To achieve these objectives, the system adopts a multi-agent architecture in which autonomous agents handle tasks such as data acquisition, context detection, model training, prediction, adaptation, and communication. The system’s effectiveness is demonstrated through five real-world building case studies, covering forecasting, model selection, system and results explanation, adaptation, retraining, and full multi-agent system orchestration. Results confirm its ability to deliver accurate, explainable, and adaptive forecasts under dynamic conditions, supporting more sustainable and trustworthy decision-making.
Rumo a uma estrutura de verificação e validação baseada em arquiteturas de microserviços
Publication . HERNÁNDEZ, LIDIA GLORIA WILSON; Pinho, Luís Miguel Rosário da Silva
In recent years, microservices architectures have become a popular approach in software development to address the challenges of scalability, maintainability, and agility of complex systems. This thesis focuses on the intersection of microservices architectures with formal verification techniques, especially in the context of safety-critical applications such as Assisted and/or Automated Driving solutions. The lack of research in this area requires a comprehensive investigation of how microservices architectures and formal verification techniques can interact effectively. The primary challenge is to seamlessly integrate microservices architectures with formal verification processes to ensure the safety properties typical of Cyber-Physical Systems, with a particular focus on autonomous driving applications. The importance of this work lies in the development of a robust framework that reconciles the decentralised nature of microservices with the requirements of formal verification. Research objectives include analysing the benefits and limitations of microservices, identifying suitable formal verification tools and microservices frameworks, designing microservices APIs, and conducting experiments for validation. Successful implementations and a comprehensive overview of formal verification methods in safety-critical systems form the basis for this research. As a proof of concept, this thesis presents the implementation of VVFramework, a microservicesbased hybrid verification and validation framework that integrates NuSMV, Z3, and an automatic Python-SMV translation pipeline and demonstrates its applicability for small case studies in safety-critical domains.