Detecção de Oscilações em Processos Industriais Baseada no Envelope Espectral

• Ano de defesa: 2018
• Autor(a) principal: Faria, Heitor Guzzo de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Engenharia Elétrica; Centro Tecnológico; UFES; Programa de Pós-Graduação em Engenharia Elétrica
• Palavras-chave em Português: Detecção de oscilações; Envelope espectral; Coeficiente de variação; Processos de fabricação; Sistemas inteligentes de controle; Engenharia elétrica; Eletrônica Industrial, Sistemas e Controles Eletrônicos; 621.3
• Link de acesso: http://repositorio.ufes.br/handle/10/9567

Oscillatory disturbance is a common issue in industrial processes. The detection and diagnosis of oscillationsis important becauseitcan indicate a fault or wear on equipment. The detection is also important because oscillation can propagate throughout the plant causing increase in variability, reduction of process performance and reduction of product quality. This thesis presents a proposal of oscillation detection based on the spectral envelope method. The automation of the methodology presented in the literature allows the implementation in industry without the requirement of a human expert to run the algorithm. Data segmentation was performed together with filtering to guarantee that disturbance free data was used in the analysis. The signals with lower impact were eliminated using the coefficient of variation.The spectrum of the remaining signals was estimated using the correlation betweensignals and the spectral envelope was calculated via the greatest eigenvalue of the estimated spectrum. Band selection was performed in an automated manner and the most relevant ones were selected using a percentage energy threshold. The oscillation indication was confirmed using a statistical hypothesis test. Signals with values over the threshold of a chi-squared distribution with 2 degrees of freedom indicate the presence of oscillation in the signal. Case studies were carried out using industrial data from an oil and gas processing platform showing that repeatability of detection results was achieved in subsequent analysis.