Klaus de Geus
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Klaus de Geus has a total of 21 co-authors. Here are the top co-authors by number of shared publications:
- Walmor Cardoso Godoi
Universidade Tecnológica Federal do Paraná
3 shared publications - Marcelo Rodrigues Bessa
Universidade Federal do Paraná, Brasil
2 shared publications A Joukoski
1 shared publicationAlexandre Augusto Angelo de Souza
Dep. of Electrical Energy and Automation Polytechnic School of São Paulo University São Paulo, Brazil
1 shared publicationAlexandre Pereira de Faria
Universidade Federal do Paraná, Brazil
1 shared publication- Daniel Henrique Marco Detzel
Universidade Federal do Paraná, Brasil
1 shared publication
Latest Publications
- Open Access
- 17 September 2024
- Published by Associação de Pesquisadores Científicos de Estudos Interdisciplinares in Energy Review: Sustainable Systems for the Future
- Vol. 39
Reconstrução Tridimensional de Corpos de Prova de Concreto via Marching Cubes Acelerado por GPU
O monitoramento da integridade de estruturas de concreto é essencial para garantir sua segurança e longevidade. Ensaios não destrutivos baseados em simulações mecânicas requerem modelos virtuais tridimensionais de alta fidelidade, extraídos de amostras reais. A tomografia industrial fornece dados volumétricos precisos, porém a geração de malhas poligonais a partir desses dados, necessária para as simulações, é computacionalmente intensiva. Este trabalho apresenta uma implementação intensamente paralela do algoritmo Marching Cubes utilizando GPU e a arquitetura CUDA/C para reconstruir isossuperfícies de corpos de prova de concreto. A solução proposta foi integrada a uma interface em C#/.NET, permitindo visualização e exportação dos modelos. Os resultados demonstram uma redução drástica no tempo de processamento: de mais de 12 horas em uma implementação serial para aproximadamente 40 segundos na versão paralela, viabilizando a utilização de modelos de alta definição em simulações mecânicas. O ganho de desempenho ocorreu com um aumento controlado no consumo de memória devido ao gerenciamento mais sofisticado dos dados.
- 17 September 2024
- Published by Associação de Pesquisadores Científicos de Estudos Interdisciplinares in Energy Review: Sustainable Systems for the Future
- Vol. 39
Open AccessReconstrução Tridimensional de Corpos de Prova de Concreto via Marching Cubes Acelerado por GPU
O monitoramento da integridade de estruturas de concreto é essencial para garantir sua segurança e longevidade. Ensaios não destrutivos baseados em simulações mecânicas requerem modelos virtuais tridimensionais de alta fidelidade, extraídos de amostras reais. A tomografia industrial fornece dados volumétricos precisos, porém a geração de malhas poligonais a partir desses dados, necessária para as simulações, é computacionalmente intensiva. Este trabalho apresenta uma implementação intensamente paralela do algoritmo Marching Cubes utilizando GPU e a arquitetura CUDA/C para reconstruir isossuperfícies de corpos de prova de concreto. A solução proposta foi integrada a uma interface em C#/.NET, permitindo visualização e exportação dos modelos. Os resultados demonstram uma redução drástica no tempo de processamento: de mais de 12 horas em uma implementação serial para aproximadamente 40 segundos na versão paralela, viabilizando a utilização de modelos de alta definição em simulações mecânicas. O ganho de desempenho ocorreu com um aumento controlado no consumo de memória devido ao gerenciamento mais sofisticado dos dados.
- Open Access
- 12 October 2023
- Published by Associação de Pesquisadores Científicos de Estudos Interdisciplinares in Energy Review: Sustainable Systems for the Future
- Vol. 38, 16-21
Segmentação de estudos tomográficos de amostras de concreto utilizando redes neurais artificiais
Em contextos como barragens de usinas hidrelétricas, a integridade das estruturas de concreto deve ser assegurada periodicamente. Para isso, laboratórios especializados realizam testes destrutivos em amostras periodicamente, o que significa que uma nova amostra é necessária para um teste adicional. Para otimizar o processo, foi idealizado um método não destrutivoutilizando tomografia industrial e a reconstrução virtual da geometria, de forma que a estrutura interna permanecesse fiel ao objeto real. No entanto, as imagens tomográficas introduzem ruído de alta frequência, o que pode dificultar o processo de reconstrução baseado no algoritmo Marching Cubes; esses artefatos são interpretados como alterações de densidade, aparecendo como minúsculos corpos a serem reconstruídos. Os métodos tradicionais de processamento de imagens falham em filtrar todos esses artefatos da imagem. Para solucionar esses desafios, uma rede neural artificial foi desenvolvida para classificar os componentes das imagens tomográficas. Para treinar a rede neural artificial, foi utilizada uma imagem segmentada, na qual o ruído de alta frequência foi filtrado e parte do ruído de baixa frequência foi preservada.
- 12 October 2023
- Published by Associação de Pesquisadores Científicos de Estudos Interdisciplinares in Energy Review: Sustainable Systems for the Future
- Vol. 38, 16-21
Open AccessSegmentação de estudos tomográficos de amostras de concreto utilizando redes neurais artificiais
Em contextos como barragens de usinas hidrelétricas, a integridade das estruturas de concreto deve ser assegurada periodicamente. Para isso, laboratórios especializados realizam testes destrutivos em amostras periodicamente, o que significa que uma nova amostra é necessária para um teste adicional. Para otimizar o processo, foi idealizado um método não destrutivoutilizando tomografia industrial e a reconstrução virtual da geometria, de forma que a estrutura interna permanecesse fiel ao objeto real. No entanto, as imagens tomográficas introduzem ruído de alta frequência, o que pode dificultar o processo de reconstrução baseado no algoritmo Marching Cubes; esses artefatos são interpretados como alterações de densidade, aparecendo como minúsculos corpos a serem reconstruídos. Os métodos tradicionais de processamento de imagens falham em filtrar todos esses artefatos da imagem. Para solucionar esses desafios, uma rede neural artificial foi desenvolvida para classificar os componentes das imagens tomográficas. Para treinar a rede neural artificial, foi utilizada uma imagem segmentada, na qual o ruído de alta frequência foi filtrado e parte do ruído de baixa frequência foi preservada.
- 30 June 2023
- Published by IGI Global Scientific Publishing
- p. 176-213
User Experience and Learning Evaluation in Virtual Training of Critical Activities
This chapter addresses methods and techniques explored in the development of a virtual environment solution for the training of electricians in critical activities, focusing on research topics, such as user experience and learning evaluation. The main concept of the virtual environment is the mapping between virtual reality technology and gamification methods with learning theories. User experience (UX) is used to assess how the system is perceived by the user. Non-player characters are modelled to assist the trainee in the learning process, focusing on decision making in adverse conditions. An automatic feedback system based on the visualization of error patterns highlights both the mistakes made in the virtual experience and the strategy for solving the proposed problem. A trainee classification process is proposed based on the analysis of human error patterns during the execution of a task. Clustering techniques applied to error patterns allow for the identification of prototypes of performance classes and their visualization in the form of distinct groups. Request access from your librarian to read this chapter's full text.
- 30 June 2023
- Published by IGI Global Scientific Publishing
- p. 176-213
User Experience and Learning Evaluation in Virtual Training of Critical Activities
This chapter addresses methods and techniques explored in the development of a virtual environment solution for the training of electricians in critical activities, focusing on research topics, such as user experience and learning evaluation. The main concept of the virtual environment is the mapping between virtual reality technology and gamification methods with learning theories. User experience (UX) is used to assess how the system is perceived by the user. Non-player characters are modelled to assist the trainee in the learning process, focusing on decision making in adverse conditions. An automatic feedback system based on the visualization of error patterns highlights both the mistakes made in the virtual experience and the strategy for solving the proposed problem. A trainee classification process is proposed based on the analysis of human error patterns during the execution of a task. Clustering techniques applied to error patterns allow for the identification of prototypes of performance classes and their visualization in the form of distinct groups. Request access from your librarian to read this chapter's full text.
- Open Access
- 1 January 2023
- Published by FapUNIFESP (SciELO) in RBRH
- Vol. 28, e46
Generation of synthetic flow scenarios by means of multivariate sampling of contemporaneous ARMA model outputs
This work constitutes one of the outcomes of the “Evaluation of hydrological scenario generation models” activity initiated by the Hydrological Scenario Representation Working Group (GT CH) and coordinated by ONS and CCEE. We introduce the LYNX-Series model, a contemporaneous non-periodic and multivariate variation of the autoregressive moving average model (CARMA) for generating synthetic time series of average inflow discharges to reservoirs in the Brazilian National Interconnected System (SIN). Notably, LYNX-Series couples the synthetic series generator with a multivariate sampling process to select a group of synthetic hydrological scenarios based on a similarity criterion with recent historical data. In addition to reducing the computational burden of the hydrothermal dispatch optimization process, the solution aims to enhance the representativeness of synthetic hydrological scenarios. The paper expounds on the theoretical aspects of the model and presents numerical simulations that validate its ability to replicate hydrological behaviors in various Brazilian basins.
- 1 January 2023
- Published by FapUNIFESP (SciELO) in RBRH
- Vol. 28, e46
Open AccessGeneration of synthetic flow scenarios by means of multivariate sampling of contemporaneous ARMA model outputs
This work constitutes one of the outcomes of the “Evaluation of hydrological scenario generation models” activity initiated by the Hydrological Scenario Representation Working Group (GT CH) and coordinated by ONS and CCEE. We introduce the LYNX-Series model, a contemporaneous non-periodic and multivariate variation of the autoregressive moving average model (CARMA) for generating synthetic time series of average inflow discharges to reservoirs in the Brazilian National Interconnected System (SIN). Notably, LYNX-Series couples the synthetic series generator with a multivariate sampling process to select a group of synthetic hydrological scenarios based on a similarity criterion with recent historical data. In addition to reducing the computational burden of the hydrothermal dispatch optimization process, the solution aims to enhance the representativeness of synthetic hydrological scenarios. The paper expounds on the theoretical aspects of the model and presents numerical simulations that validate its ability to replicate hydrological behaviors in various Brazilian basins.
- 18 August 2021
- Published by Springer Nature in Journal of Control, Automation and Electrical Systems
- Vol. 32 (6) , 1704-1717
Optimization Criteria Weighing in Dynamic Optimal Power Flow Performing Monthly, Weekly and Daily Hydrothermal Generation Dispatch
- 18 August 2021
- Published by Springer Nature in Journal of Control, Automation and Electrical Systems
- Vol. 32 (6) , 1704-1717
Optimization Criteria Weighing in Dynamic Optimal Power Flow Performing Monthly, Weekly and Daily Hydrothermal Generation Dispatch
- 1 April 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 4, 1-7
Impact of distributed generation on the operational planning of low voltage distribution networks using genetic algorithms
This paper presents a study of the impact of Distributed Generation (DG) in the planning and design of Low Voltage (LV) distribution networks. The proposed model considers the planning of LV networks with the aims of reducing power losses, acceptable voltage levels and maximizing power supply of distributed generation units using Genetic Algorithm (GA). Planning LV distribution networks involves radial configuration design, cable replacement, changes on the phases of the consumers and transformer position. In this work, these goals are considered by taking into account different conductors, voltage drop and power losses in lines. A power flow for radial distribution networks based on a backward/forward sweep using current summation was presented to validate the solutions.
- 1 April 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 4, 1-7
Impact of distributed generation on the operational planning of low voltage distribution networks using genetic algorithms
This paper presents a study of the impact of Distributed Generation (DG) in the planning and design of Low Voltage (LV) distribution networks. The proposed model considers the planning of LV networks with the aims of reducing power losses, acceptable voltage levels and maximizing power supply of distributed generation units using Genetic Algorithm (GA). Planning LV distribution networks involves radial configuration design, cable replacement, changes on the phases of the consumers and transformer position. In this work, these goals are considered by taking into account different conductors, voltage drop and power losses in lines. A power flow for radial distribution networks based on a backward/forward sweep using current summation was presented to validate the solutions.
- 1 October 2010
- Published by British Institute of Non-Destructive Testing (BINDT) in Insight - Non-Destructive Testing and Condition Monitoring
- Vol. 52 (10) , 533-539
Automated flaw detection in X-ray tomography of polymer insulators
- 1 October 2010
- Published by British Institute of Non-Destructive Testing (BINDT) in Insight - Non-Destructive Testing and Condition Monitoring
- Vol. 52 (10) , 533-539
Automated flaw detection in X-ray tomography of polymer insulators
- 1 December 2009
- Published by British Institute of Non-Destructive Testing (BINDT) in Insight - Non-Destructive Testing and Condition Monitoring
- Vol. 51 (12) , 654-659
Determination of crushed stone volume in concrete cores from hydroelectric power plant dams by three-dimensional tomography
In this paper we present the results of a methodology for obtaining the volume of crushed stones using industrial tomography in concrete cores from hydroelectric power plant dams. In order to apply the appropriate voltage in the X-ray tube during radiograph acquisition, the average linear attenuation coefficients () of the cement paste and the mortar were determined. The two-dimensional tomographic images were reconstructed using 180 digital X-ray projections obtained by an industrial tomograph. The 3D tomography images were rendered using slices in the bitmap format and the VTK (Visualization Toolkit) open source library, and finally, the Marching Cubes algorithm was used for the volumetric visualisation of the data. The low contrast resolution was also evaluated. The methodology for volume determination was evaluated using two phantoms (one with cement paste and another with mortar).
- 1 December 2009
- Published by British Institute of Non-Destructive Testing (BINDT) in Insight - Non-Destructive Testing and Condition Monitoring
- Vol. 51 (12) , 654-659
Determination of crushed stone volume in concrete cores from hydroelectric power plant dams by three-dimensional tomography
In this paper we present the results of a methodology for obtaining the volume of crushed stones using industrial tomography in concrete cores from hydroelectric power plant dams. In order to apply the appropriate voltage in the X-ray tube during radiograph acquisition, the average linear attenuation coefficients () of the cement paste and the mortar were determined. The two-dimensional tomographic images were reconstructed using 180 digital X-ray projections obtained by an industrial tomograph. The 3D tomography images were rendered using slices in the bitmap format and the VTK (Visualization Toolkit) open source library, and finally, the Marching Cubes algorithm was used for the volumetric visualisation of the data. The low contrast resolution was also evaluated. The methodology for volume determination was evaluated using two phantoms (one with cement paste and another with mortar).