CRCassius Zanetti Resende

Cassius Zanetti Resende

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Cassius Zanetti Resende has a total of 18 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    PREVISÃO DE GERAÇÃO DE ENERGIA EÓLICA E SOLAR UTILIZANDO REDES NEURAIS RECORRENTES E CONVOLUCIONAIS: UMA ABORDAGEM HÍBRIDA CNN-GRU

    by Fábio Pinto Monte, Fidelis Zanetti de Castro, Cassius Zanetti Resende

    Este artigo expande um trabalho anterior que investigou a previsão de geração de energia eólica e solar utilizando redes neurais recorrentes (LSTM e GRU). Reconhecendo a importância da precisão para a integração de energias renováveis na rede elétrica, o presente estudo aprimora a metodologia original ao incorporar uma camada de Rede Neural Convolucional (CNN) a um modelo GRU, criando uma arquitetura híbrida CNN-GRU. O objetivo é explorar como a CNN pode melhorar a extração de características de dados meteorológicos, otimizando a previsão. A metodologia envolveu a coleta e pré-processamento de dados meteorológicos e de geração de energia do IEEE DataPort, seguida pela construção, treinamento e avaliação dos modelos LSTM, GRU e CNN-GRU utilizando validação cruzada temporal. Os resultados demonstraram que, embora o LSTM tenha apresentado o melhor desempenho médio (R² de 0.946 ± 0.005), o modelo híbrido CNN-GRU (R² de 0.925 ± 0.008) superou o GRU autônomo (R² de 0.917 ± 0.008). Esta melhoria sugere que a camada CNN foi eficaz na identificação de padrões locais nos dados meteorológicos, contribuindo para previsões mais robustas. O estudo conclui que a abordagem híbrida CNN-GRU é promissora para aprimorar a previsão de energia renovável, especialmente em cenários com dados de entrada complexos.

  • Open Access

    High‐Fidelity Synthetic Raman Spectra Generation for Sinter Basicity Prediction Using β‐Variational Autoencoders

    by Marjorie Ariele Pereira, Daniel Cruz Cavalieri, Adilson Ribeiro Prado, Cassius Zanetti Resende, Erica Simões Rodrigues

    Raman spectroscopy has emerged as a powerful analytical tool across diverse industrial sectors, owing to its nondestructive nature, high chemical specificity, and ability to provide unique molecular “fingerprints.” In the steelmaking industry, this technique offers a promising route for the rapid and precise characterization of critical materials such as sinter, a porous agglomerate of iron ore fines essential for blast furnace charging. However, practical deployment is often hindered by the scarcity of labeled spectral data. This work investigates the use of a ‐variational autoencoder (‐VAE) for the generation of high‐fidelity synthetic Raman spectra from a limited set of real sinter samples (321 spectra). With and latent codes sampled from a full‐covariance Gaussian fitted to the aggregate posterior, the model produces genuinely novel spectra whose global distributional fidelity matches that of SMOTE (Fréchet spectral distance: 4.55 vs. 4.41), despite SMOTE being constructed by interpolation of existing spectra; the mean intensity correlation with real spectra is 0.930 0.038. These synthetic spectra were used to augment supervised regression models for basicity () prediction. The LightGBM model achieved (95% CI: [0.72, 0.84]) in cross‐validation with original data; no augmentation method produced a statistically significant improvement (Wilcoxon signed‐rank, all ), and on the held‐out test set (), no augmented model exceeded the original‐only baseline (; S‐VAE: 0.809; Interpolação: 0.786; ‐VAE: 0.780; SMOTE: 0.779). The operational value of generative augmentation emerges instead in a downstream classification experiment: classifiers trained with increasing fractions of synthetic spectra sustain higher macro F1 at 100% synthetic fraction (‐VAE 0.66, S‐VAE 0.60) than SMOTE‐augmented classifiers, which degrade to 0.58. These results establish that the value of VAE‐based generation lies in producing distributionally faithful yet novel spectra that remain useful when consumed directly by downstream tasks such as spectral library expansion and classifier training, rather than in regression gains.

  • Open Access

    Development of Perception and Control Applications for Collaborative Robots Using ROS2

    by Anonione da Silva Mascarenhas Pinho, Cassius Zanetti Resende, Daniel Cruz Cavalieri

    This work explores the integration of perception systems in behavior-based control applications using ROS2, with a focus on safe and effective interactions between humans and collaborative robots. Unlike conventional industrial robots, collaborative robots are designed to interact directly with human operators without physical barriers, making automation more accessible, particularly for small and medium-sized enterprises. A system integrating the UR3e robot and the Intel RealSense D435i camera was developed, enabling the robot to recognize objects and interpret gestures. YOLOv8 was used for object detection and segmentation, MediaPipe for gesture recognition and face detection, and Detectron2 for human detection, empowering the robot to manipulate objects and respond to gesture commands. The system's efficiency was validated through practical tests by moving objects between known points. The study highlights opportunities for improvement, such as replacing Detectron2 with YOLOv8, expanding the dataset, and adopting low-latency kernels. This research contributes to collaborative robotics by demonstrating the feasibility of integrating perception and behavior-based control, laying the foundation for future investigations and applications.

  • Open Access

    Checking the Status of High Voltage Disconnection Switches Using Siamese Convolutional Neural Networks

    by Celso Soares Godoy, Marcelo Denadai Marcon, Gustavo Maia de Almeida, Daniel Cruz Cavalieri, Cassius Zanetti Resende

    A high voltage power substation is an electrical installation made up of equipment responsible for transmission and distribution of energy for voltages above 69kV and below 230kV. Necessary to keeping our homes and industries running, a substation has various power equipment such as circuits breakers, power transformers and disconnection switches that guarantee its operation. These devices can be controlled locally or remotely way, either through control systems or manually by a human operator.One of the fundamental pieces of equipment to guarantee the operation of a substation is a disconnection switch. Disconnection switches are electromechanical equipment composed by moving parts, in a conventional substation. This equipment is subject to mechanical efforts, vibration and temperature. The wrong functioning of a disconnection switch can cause major losses for the company, from equipment downtime to a complete lack of power to the plant.With the aim of contributing to avoiding failures in the operation of an industrial substation, this work proposes the use of artificial intelligence (AI) and machine learning techniques, through computer vision based on the Siamese convolutional neural network to verify the disconnection switches opened and closed status of a power substation in a steel industry.

  • Diagnóstico de Falhas em Rolamentos de Motores Elétricos com Base na Análise da Assinatura da Corrente do Motor

    by Marcos Romulo Oliveira, Luiz Alberto Pinto, Cassius Resende

    Título: Diagnóstico de Falhas em Rolamentos de Motores Elétricos com Base na Análise da Assinatura da Corrente do Motor Autores: Marcos Romulo de Oliveira, Luiz Alberto Pinto, Cassius Zanetti Resende Resumo: This article presents a comparative analysis of the performance Read More ...

  • Improving sEMG-Based Hand Gesture Recognition Using Maximal Overlap Discrete Wavelet Transform and an Autoencoder Neural Network

    by Fernando Henrique Cruz de Andrade, Flavio Garcia Pereira, Cassius Zanetti Resende, Daniel Cruz Cavalieri


  • Estimation and prediction of motor load torque applied to electrical submersible pumps

    by André Paulo Ferreira Machado, Cassius Zanetti Resende, Daniel Cruz Cavalieri