https://tede.unioeste.br/handle/tede/550 Thu, 18 Jun 2026 04:31:33 GMT 2026-06-18T04:31:33Z https://tede.unioeste.br/handle/tede/8468 Title: Recomendação de quantificadores utilizando meta-aprendizado Autor: Gomes, Guilherme Bachega Primeiro orientador: Maletzke, André Gustavo Abstract: Metalearning based recommender systems have become an effective approach for the automatic selection of algorithms in machine learning tasks. This approach has been successfully employed in several studies to build recommenders across different domains. In the literature, there are works that investigate the recommendation of classifiers, hyperparameter configurations, and feature selection algorithms. Although a wide range of recommendation approaches has been proposed for classification problems, there is a lack of work in the context of quantification. Recently, quantification has attracted increasing interest in the machine learning community. Unlike classification, whose goal is to assign labels to unlabeled instances, quantification aims to estimate the class distribution in an unlabeled dataset. Several quantification methods have been proposed over the last years and, as in classification, there is no quantifier that consistently outperforms the others across all scenarios. Moreover, the empirical evaluation of quantifiers is particularly costly, as it requires generating multiple test sets with controlled prevalences. In this setting, determining the best quantifier for a given dataset is critical, yet labor-intensive when based on trial-and-error strategies. To address this gap, this work proposes the first meta-learning architecture for quantifier recommendation. The architecture models the relationship between dataset characteristics and quantifier performance to recommend a ranking of the most suitable quantifiers, without requiring trial-and-error experimentation. To evaluate the architecture, two quantifier recommendation strategies are developed. Experimental results demonstrate its effectiveness: under the best strategy, the system places the best-performing quantifier at the top of the recommended ranking in 91 out of 100 evaluated datasets. In addition, the ranking is directly explored in quantification tasks. When applying only the top-ranked quantifier, the method achieves performance comparable to the state-of-the-art DyS quantifier across the 100 datasets. On more complex datasets, performance remains stable and comparable to the best quantifier. Finally, combining the top quantifiers in the recommended ranking yields performance comparable to the state of the art over all datasets and superior performance when considering only the more complex datasets. Publisher: Universidade Estadual do Oeste do Paraná Tipo do documento: Dissertação Fri, 06 Mar 2026 00:00:00 GMT https://tede.unioeste.br/handle/tede/8468 2026-03-06T00:00:00Z https://tede.unioeste.br/handle/tede/8290 Title: Influência de técnicas de sincronização de conversores na distorção harmônica de microrredes Autor: Bender, Samuel Graff Primeiro orientador: Almeida, Adriano Batista de Abstract: Microgrids (MGs) are smaller-scale power systems used to integrate distributed energy resources into the main electrical grid through power converters. To ensure safe and efficient operation of these systems, both in grid-connected and islanded modes, control strategies are essential for power sharing and maintaining voltage and frequency stability. In the context of converter control, synchronization techniques, such as Phase-Locked Loops (PLLs), are employed to estimate the grid frequency and phase, ensuring the converters are synchronized with the voltage signal at their point of connection. However, these control systems are vulnerable to power quality issues, mainly voltage imbalances and harmonic distortions in the voltage signals. The presence of converters and non-linear loads are some of the causes of harmonic distortions in MGs, negatively affecting the performance of synchronization techniques. Based on the hypothesis that synchronization techniques may affect the harmonic distortion of voltage and current signals, and given the scarcity of studies regarding this problem, the present work aims to fill this gap in the literature by evaluating the dynamic behavior of synchronization techniques and their impact on the harmonic distortion of MGs. Four synchronization techniques were selected based on the literature and implemented on Digital Signal Processors (DSPs). The techniques were subjected to two types of tests: the first involved applying known signals using the UNIOESTE MG Laboratory structure, where their dynamic performance was evaluated under events of frequency and phase variation, and harmonic distortion. The second type of test was conducted via simulations in PSIM software, in which the laboratory MG model was implemented. The simulations considered the connection of a non-linear load model at different operating points and under various converter combinations. The experimental results showed that synchronization techniques with advanced filtering capabilities have better dynamic performance, ensuring robustness against grid disturbances. In the simulations, it was concluded that there is no significant coupling between the chosen synchronization technique and the harmonic distortion of the electrical quantities in the evaluated scenarios. The main contribution of this work is the verification of the hypothesis in the MG context, in addition to the development of the laboratory MG structure and the detailed computational model for future studies. Publisher: Universidade Estadual do Oeste do Paraná Tipo do documento: Dissertação Fri, 10 Oct 2025 00:00:00 GMT https://tede.unioeste.br/handle/tede/8290 2025-10-10T00:00:00Z https://tede.unioeste.br/handle/tede/8210 Title: Comparação de protótipo localizador de faltas com relés comerciais: um estudo para identificação de melhorias Autor: Pereira, Eduardo Henrique Primeiro orientador: Motter, Daniel Abstract: Fault location (FL) in transmission lines (TL) plays a strategic role in the operation of electric power systems, as it makes it possible to precisely identify the point of occurrence of disturbances and reduce system restoration time. In this way, accurate FL estimates contribute to efficient service restoration and optimize maintenance activities, thus increasing system reliability. With the objective of monitoring signals and recording disturbances at points of interest, Itaipu Parquetec, with the support of ITAIPU Binacional, developed a device that integrates the functionalities of a Digital Fault Recorder (DFR) and a Phasor Measurement Unit (PMU), resulting in the Disturbance Recorder and Phasor Measurement Unit (RPMF), with its latest update being the addition of an LF module for LTs. Considering the need to test and enhance these updates, this work presents a comparative evaluation of the RPMF’s LF prototype with commercial LF estimators, in order to assess the technological compatibility between the products, as well as to verify whether the proposed multi-method can be applied to more critical LTs without the need for flexibility. To achieve this, real-time simulations were carried out using a Hardware-inthe-loop configuration, in which the RPMF and two intelligent electronic devices (IEDs) were connected in parallel to amplifiers controlled by a Real-Time Digital Simulator (RTDS). The LTused as a model in the tests is part of the Paraguayan power system, selected due to ITAIPU’s interest and aiming for future applications of the LF prototype at the location. In addition to the comparison with the commercial IEDs, the performance of the main LF methods present in the RPMF algorithm was also evaluated, namely: (i) Impedance Method, (ii) Reactance Method, (iii) Simple Takagi, (iv) Negative-Sequence Polarized Takagi, and (v) Zero-Sequence Polarized Takagi. A series of tests was conducted simulating various fault scenarios, including variations in fault type, resistance, incidence angle, and distance to the fault point. The analysis of the results allowed for a comparison between the methods implemented in the RPMF and the other evaluated equipment. The results obtained demonstrate that the RPMF presents satisfactory performance, comparable to the commercial solutions analyzed, and stands out in various scenarios. Furthermore, the analysis of the RPMF methods evidenced potential for multimethod improvement, revealing the possibility of recommendations for strategic adjustments that would increase its accuracy in more critical LTs. Thus, the study reinforces the viability of the RPMF as a promising national solution for LF, distinguishing itself by its performance and flexibility for adaptation to different conditions, while also indicating paths for future developments. Publisher: Universidade Estadual do Oeste do Paraná Tipo do documento: Dissertação Thu, 27 Nov 2025 00:00:00 GMT https://tede.unioeste.br/handle/tede/8210 2025-11-27T00:00:00Z https://tede.unioeste.br/handle/tede/8208 Title: Avaliação de representações simbólicas de movimentos articulares humanos com SAX e SFA a partir de dados de acelerômetro Autor: Merencia, Ana Paula Primeiro orientador: Lee, Huei Diana Abstract: The monitoring and analysis of human joint movements using inertial sensors have become accessible and effective alternatives in fields such as rehabilitation, ergonomics, and telemedicine. However, the high dimensionality and individual variability of these signals hinder direct interpretation, calling for representation strategies that reduce complexity without discarding relevant information. This study investigates the symbolic representation methods SAX and SFA applied to time series acquired by smartphone accelerometers from eight healthy participants during the controlled execution of elbow flexion/extension, shoulder abduction/adduction, and shoulder circumduction. The experimental protocol comprised three bouts of ten cycles for each movement, with standardization and supervision by a physical education professional. We evaluated different symbolic configurations by varying the alphabet size (3, 7, and 10 symbols) and the strategies for generating the reference word, and performed a quantitative analysis based on the Hamming distance between symbolic sequences. Statistical significance was assessed using the Friedman test with Nemenyi post-hoc multiple comparisons. Results show that Standard SFA yielded, in most configurations, lower Hamming distances and greater between-subject homogeneity—especially with reduced alphabets (três and sete symbols)—confirming its suitability for more regular movements such as elbow flexion/extension and shoulder abduction/adduction. For shoulder circumduction, characterized by higher inter- and intra-subject variability, Standard SAX performed better with smaller alphabets, underscoring the importance of selecting the method according to movement complexity. Overall, the symbolic approaches preserved salient signal patterns at low computational cost, supporting their potential for deployment in rehabilitation systems and telemedicine e-health platforms. This work contributes a detailed comparative analysis between SAX and SFA across different movements and conditions, highlighting their limitations, strengths, and scenarios of best applicability. Publisher: Universidade Estadual do Oeste do Paraná Tipo do documento: Dissertação Thu, 02 Oct 2025 00:00:00 GMT https://tede.unioeste.br/handle/tede/8208 2025-10-02T00:00:00Z