Discrete-time domain poles zeros identification using back propagation neural networks

Research output: Journal Publications and ReviewsRGC 22 - Publication in policy or professional journal

View graph of relations



Original languageEnglish
Pages (from-to)215-218
Journal / PublicationIEE Conference Publication
Publication statusPublished - 1991


Title6th International Conference on Digital Processing of Signals in Communications
PlaceUnited Kingdom
Period2 - 6 September 1991


This paper describes a back propagation neural network applying to poles zeros identification in discrete time domain. Traditional Recursive Least Squares (RLS) algorithm is time consuming and sensitive to noise. Neural networks possess massive parallel processing capability and noise immunity, the time and noise constraints can be eliminated. The results are encouraging and demonstrate that neural networks offer new promising directions towards solving system identification problems radically.