Uso de Redes Neurais na Desconvolução Preditiva de Traços Sísmicos

Deconvolution of mixed phase wavelets is a challenging problem which is not successfully solved through linear predictors. Thus, this paper investigates the use of nonlinear Forward Prediction Error Filters on mixed phase scenarios. This approach is based on the effectiveness of theses structures in mixed phase equalization problems and the preliminary results are promising, which encourage further research in more complex scenarios.