You are in the accessibility menu

Please use this identifier to cite or link to this item: http://acervodigital.unesp.br/handle/11449/87157
Title: 
Momentos nulos e regularidade wavelet na detecção de falhas em sinais
Author(s): 
Uzinski, Julio Cezar
Institution: 
Universidade Estadual Paulista (UNESP)
Abstract: 
  • Geralmente, em processamento digital de sinais, quando um sinal vai ser analisado usando a transformada wavelet, algumas ou várias funções wavelets são testadas e aquela que melhor se adapta no tempo e na frequência com o dado sinal é a escolhida. Entretanto, sabese que propriedades, como regularidade, suavidade e suporte compacto são determinantes na escolha da função wavelet a ser usada. Funções com suporte maior aumentam a resolução em frequência, mas diminuem a resolução temporal do sinal transformado. Assim, deve-se escolher uma função wavelet cujo suporte não comprometa nenhuma das resoluções. Momentos nulos e suavidade estão matematicamente relacionados, pois quanto maior o número de momentos nulos de uma wavelet, mais suave ela é. Quanto maior a suavidade da wavelet, maior a probabilidade de reconstrução perfeita do sinal decomposto pela transformada wavelet. Segundo a formulação de Sherlock & Monro o espaço de wavelets ortonormais é parametrizado por um conjunto de parâmetros angulares, adaptando o trabalho com matrizes paraunitárias. Esta parametrização pode ser usada para ajustar a wavelet, a fim de melhorar o índice de desempenho de algumas aplicações relacionadas ao processamento de sinais em questão. A formulação de Sherlock & Monro fora estendida para até dois momentos nulos em trabalhos anteriores e, respectivamente, para três momentos nulos neste trabalho. Além da extensão dessa formulação, este texto apresenta uma aplicação da mesma em relação ao diagnóstico de falhas em sinais elétricos.
  • Generally, in digital signal processing when a signal is to be analyzed using the wavelet transform, some or several wavelet functions are tested and the one that best matches in time and frequency with the given signal is chosen. However, it is known that properties such as regularity, smoothness and compact support are decisive in the choice of the wavelet function to be used. Functions with greater support increase frequency resolution, on the other hand decrease the temporal resolution of the transformed signal. Thus, one should pick a wavelet function which support does not compromise any of the resolutions. Vanishing Moments and smoothness are closely connected, since the greater is the number of vanishing moments the smoother is the wavelet. And how much smoother is the wavelet, the greater is the probability of perfect reconstruction of the signal decomposed by wavelet transform. According to the formulation of Sherlock and Monro the space of orthonormal wavelets is parameterized by a set of angular parameters, adapting the work of factorization of paraunitary matrices. That parameterization could be used to adjust the wavelet in order to improve the performance index of some applications related to signal processing. The formulation of Sherlock and Monro was extended for up to two vanishing moments in previous papers and, respectively, and to up to three vanishing moments in this work. Beyond the extent of this formulation, this text presents an application of the constructed wavelets for the diagnosis of faults in electrical signals.
Issue Date: 
23-Aug-2013
Citation: 
UZINSKI, Julio Cezar. Momentos nulos e regularidade wavelet na detecção de falhas em sinais. 2013. 103 f. Dissertação (mestrado) - Universidade Estadual Paulista Júlio de Mesquita Filho, Faculdade de Engenharia de Ilha Solteira, 2013.
Time Duration: 
103 f.
Publisher: 
Universidade Estadual Paulista (UNESP)
Keywords: 
  • Wavelets (Matematica)
  • Teoria dos sinais (Telecomunicações)
  • Automation
URI: 
Access Rights: 
Acesso aberto
Type: 
outro
Source:
http://repositorio.unesp.br/handle/11449/87157
Appears in Collections:Artigos, TCCs, Teses e Dissertações da Unesp

There are no files associated with this item.
 

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.