https://repositorio.ufu.br/handle/123456789/5466 2026-09-12T01:36:42Z https://repositorio.ufu.br/handle/123456789/50156 Title: Detecção de anomalias nas respostas de vibração de rotores usando técnicas de inteligência artificial Abstract: Rotating machinery plays an essential role in industrial sectors such as oil and gas, power generation and mining, and its failures may result in unplanned shutdowns, significant economic losses and safety risks. Vibration analysis stands out as one of the main condition monitoring techniques, enabling the early identification of mechanical problems and supporting predictive maintenance strategies. The increasing adoption of continuous monitoring systems has led to large volumes of data, imposing limitations on traditional approaches based exclusively on manual analysis. This dissertation proposes and evaluates Machine Learning methodologies for anomaly detection in rotor vibration responses, with emphasis on unsupervised strategies aligned with the industrial reality of scarce labeled fault data. The work integrates and extends the Edge Analytics framework, developed by the Structural Mechanics Laboratory of the Federal University of Uberlândia (LMEst - UFU), which combines simple statistical techniques and deep learning models for anomaly detection. In addition to the original models of the Edge Analytics architecture—Gates, Autoencoder, DeepAnT, and Histogram — new models based on Multilayer Perceptron (MLP) and Long Short-Term Memory (LSTM) are developed and evaluated. Furthermore, the combination of these models with Change Point Detection (CPD) techniques is investigated in order to enhance sensitivity in identifying subtle and sustained changes in vibration responses. The validation of the proposed methodologies is conducted through five benchmarks, involving numerical and experimental data from different rotating systems, including rotors supported by magnetic bearings, an industrial exhaust fan and bearing degradation tests. Performance is assessed using classical classification metrics, with emphasis on the F1 score due to the strong class imbalance. The results indicate that there is no universally superior model, however LSTMand MLP-based models stand out in faults with progressive evolution, while the combination with CPD contributes to the detection of subtle changes. Thus, the study provides technical support for industrial applications in predictive maintenance. 2026-08-08T00:00:00Z https://repositorio.ufu.br/handle/123456789/49944 Title: Manufatura aditiva de parede fina de aço inoxidável duplex via deposição a arco com e sem resfriamento ativo Abstract: Metal Additive Manufacturing (AM) has been consolidated as a strategic technology for the production of complex components, promoting production efficiency and cost reduction compared to conventional processes. Among its techniques, the Arc Deposition Additive Manufacturing (WAAM) process stands out for its high deposition rate and applicability in large parts. Thermal control during metal AM is even more relevant in thin-walled components, where steep thermal gradients and high localized temperatures can generate defects such as distortions, high residual stresses, and mechanical anisotropy. In this context, cooling techniques play a crucial role. In this context, the present study aimed to perform a comparative study between the Arc Deposition Additive Manufacturing of 2209 duplex stainless steel deposited by pulsed current using active cooling in water with natural air cooling. Duplex stainless steel was selected due to its two-phase ferrite/austenite microstructure, which provides an excellent combination of mechanical and corrosion resistance, which is widely required in critical industrial applications. The methodology involved the deposition of thin walls of duplex stainless steel by additive manufacturing via MADA process with pulsed current, using solid wire and commercially pure argon shielding gas. Two cooling conditions were compared: natural air cooling, similar to conventional welding processes, and near-immersion active cooling (NIAC), which consists of keeping the lower layers partially submerged in water or an aqueous mixture in order to reduce heat input. The samples produced underwent microstructural analyses, mechanical tests (tensile, bending, Charpy-V impact and hardness) and chemical characterizations by EDS and GDS. The results indicated that both cooling conditions allowed the fabrication of thin walls with good structural integrity and interlayer cohesion. The NIAC process showed better microstructural homogeneity, lower dispersion of mechanical results and reduced anisotropy, while natural cooling exhibited slightly higher hardness due to thermal heterogeneity. The chemical analysis confirmed the stability of the alloy and the preservation of the PREN (Pitting Resistance Equivalent Number) index, ensuring resistance to localized corrosion. It is concluded that MADA with pulsed current is a viable alternative for the manufacture of duplex stainless-steel components, and that active cooling by near-immersion represents an efficient strategy to improve thermal control, reduce residual stresses and increase process reliability. 2025-11-28T00:00:00Z https://repositorio.ufu.br/handle/123456789/49810 Title: Desenvolvimento de equipamento acoplado a jogos sérios para reabilitação da escoliose Abstract: This dissertation presents the development and evaluation of technological resources to support Physiotherapeutic Scoliosis-Specific Exercises, including two serious games, body tracking systems, and a physical support device. The proposal aims to make exercise practice more interactive, traceable, and reproducible, without replacing physiotherapeutic assessment. Initially, a literature review was carried out on Adolescent Idiopathic Scoliosis, serious games, body tracking tools, and applications related to spine rehabilitation. Next, the games Knight Run and Paint Maker were developed in Unity, using C# programming, including profile management, exercise selection, performance recording, series configuration, and data export. Game control was implemented through two different strategies, one based on Nuitrack, using the Intel RealSense D435 depth camera, and the other based on MediaPipe, using the Logitech C270 HD 720p webcam (30 FPS). An adaptation for Android devices was also developed, restricted to the use of MediaPipe. The users’ perception of the applications was evaluated with 10 participants, using adapted forms based on the Game Experience Questionnaire and the Intrinsic Motivation Inventory, resulting in overall scores of 73.12% for game experience and 83.33% for perceived motivation and usefulness. In parallel, a modular device was designed to provide support, upper limb leveling, scapular support, and fixation points for elastic resistance during the exercises. Finally, an experimental stage was carried out comparing Nuitrack and MediaPipe with the Qualisys system, adopted as the reference, during movements corresponding to the game controllers. The analysis considered repetitions, total time, regularity, angles, and depth, when applicable. The results indicated 0.00% error in repetition counting, temporal errors ranging from 0.21% to 7.60%, regularity errors predominantly below 8.33%, depth errors ranging from 0.27% to 1.04% between Nuitrack and Qualisys, and angular errors below 7.88% in most conditions, with greater deviations in frontal elevation trials using MediaPipe. It is concluded that the developed resources showed potential as complementary, accessible, and integrated tools to support scoliosis treatment, although future studies with patients, larger samples, and clinical validation are still required. 2026-08-14T00:00:00Z https://repositorio.ufu.br/handle/123456789/49588 Title: Desenvolvimento de módulos robóticos para instalação de amortecedores de vibração em cabos de alta tensão Abstract: The application of robots in high-voltage power line maintenance operations has gradually advanced over the past decades, with notable emphasis on mobile robot systems. However, despite the progress achieved, many challenges related to the development of such robots persist. Among the desirable features for these robots are: reliability, autonomous operation, low weight, high obstacle-transposing capability, and shielding against electromagnetic interference, many of which remain insufficiently addressed in existing robotic systems. Considering this, the present dissertation proposes the development of robotic modules for the installation of vibration dampers on high-voltage power lines. These robotic modules were designed to perform specific maintenance tasks in energized and hard-to-reach environments. This work describes the design, development, construction, and experimental testing of two robotic modules: one for the installation of Stockbridge-type dampers, and another for the installation of preformed-type vibration dampers. The dissertation also presents a shielding procedure for the robotic modules against electromagnetic interference and electrostatic discharges originating from transmission lines of up to 138 kV. Field tests carried out at the facilities of a power utility company demonstrated the modules’ effectiveness in overcoming a significant portion of existing limitations, positioning the developed solutions as promising alternatives in the field of high-voltage line maintenance robotics. 2026-02-23T00:00:00Z