Análise e extração de características de imagens termográficas utilizando componentes principais

• Ano de defesa: 2010
• Autor(a) principal: Santos, Gilnete Leite dos [UNESP]
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual Paulista (Unesp)
• Palavras-chave em Português: Termografia; Análise de componentes principais; Reconhecimento de padrões; Processamento de imagens; Manutenção preditiva; Análise de imagens térmicas; Classificação de padrões; Thermography; Analysis of principal component; Predictive maintenance; Pattern recognition and classification
• Link de acesso: http://hdl.handle.net/11449/108383

Thermography techniques are currently gaining ground as predictive maintenance techniques, mainly due to its non-invasive character (non-contact tool) that allows the monitoring of heating condition of machines and equipment also in operation and even energized. The use of thermographic cameras is now a reality in many industrial and electrical sectors for monitoring and fault detection based on temperature. However, the use of thermal imagers in the maintenance should not be restricted to only the evaluation of temperature, since the thermographic images are signs that show complex patterns and they can capture the different characteristics of the actual condition of the monitored equipment. Information other than temperature could be observed for a more consistent evaluation of its state of operation. This paper discusses the propose of use of the multivariate analysis technique, Principal Component Analysis (PCA) for the processing and analysis of a series of thermographic images in order to identify patterns associated with temperature variations of the images, as well as, the interpretation of these data in terms of their spatial/temporal variability. Initially the technique was used to the analysis of data (thermal images) obtained from the simulation of heating conditions of a component (electric switch) aiming at to test and verify the validity of the proposal and program development. Later the technique was applied to the monitoring and evaluation of the heating condition of components of a simplified model of a telephone panel, formed by aluminum blocks fixed in a plate of acrylic. The analysis in the spatial and temporal mode of the set of thermograms obtained for different heating conditions of the blocks, it showed that it is possible to verify and establish correlations between the Principal Components and the thermal profile of the system