Processing of impedance data records using artificial neural networks

Xavier Muñoz-Berbel, Manel Del Valle, Montserrat Cortina-Puig

Research output: Chapter in BookChapterResearchpeer-review


Impedance Spectroscopy (IS) is a highly informative technique based on the response of the system under study to the application of an AC potential of small magnitude. Complex systems generate complex responses that cannot be directly interpreted but they require a mathematical treatment. Traditionally, IS data have been analysed in terms of equivalent electrical circuit. Hence, impedance records are fitted using an equivalent circuit containing the most suitable combination of electrical elements (basically resistances and capacitances) to model the system under study. In the circuit, each electrical element corresponds to a single event thus turning a complex system into simple processes that can be individually analysed. Although most of the authors fit impedance data using variations/simplifications of the Randles' equivalent circuit, this approach usually fails when considering very complex systems. Chemometric tools, and particularly Artificial Neural Networks (ANNs), are an excellent alternative to the currently used mathematical methods. Last years, impedimetric transduction and ANN processing have been combined to obtain highly informative data. Corrosion studies, clinical diagnosis, environmental monitoring and food and beverage industrial analysis are examples of areas where IS and ANN have been successfully applied. In most of cases, IS data are analysed qualitatively by the ANN and the samples are classified in groups of data or populations. Most of the ANNs used for qualitative analysis typically need a single hidden layer. However, some authors have been using more complex ANNs to be able to get quantitative information from the IS, for example with ANNs using two hidden layers to model the data. © 2011 by Nova Science Publishers, Inc. All rights reserved.
Original languageEnglish
Title of host publicationArtificial Neural Networks
Place of PublicationNova York (US)
Number of pages29
Publication statusPublished - 1 Jan 2011


Dive into the research topics of 'Processing of impedance data records using artificial neural networks'. Together they form a unique fingerprint.

Cite this