Accession Number : ADA304240
Title : Feasibility of Using Artificial Neural Networks with Electrochemical Impedance Spectroscopy Data From Coated Steel.
Descriptive Note : Final rept.,
Corporate Author : NAVAL SURFACE WARFARE CENTER CARDEROCK DIV BETHESDA MD
Personal Author(s) : Hack, Harvey P. ; Matteson, M. A.
PDF Url : ADA304240
Report Date : DEC 1995
Pagination or Media Count : 30
Abstract : Electrochemical impedance spectroscopy (E.I.S.) techniques can provide information about the condition of protective coatings on steel marine structures. Currently, an expert is required to interpret the data produced from an E.I.S. measurement, classifying the coating as 'good' or 'poor' or identifying the data as 'bad.' This limits the use of E.I.S. techniques to experienced operators. If the E.I.S. technique is to be used for production by inexperienced operators, measurements must be classified automatically. This investigation uses artificial neural networks (ANN) to develop an automated E.I.S. data classifier. ANNs were trained with a large database of measurements oil known good or poor coatings, including some bad data. The ANNs were tested with E.I.S. data not included in the training set. A variety of measurement signal processing schemes and network structures was evaluated. ANNs were developed which can accurately determine if the coating is good or poor and whether measurement problems produced bad data. (MM)
Descriptors : *NEURAL NETS, *STEEL, *PROTECTIVE COATINGS, COMPUTERIZED SIMULATION, SHIPS, SPECTROSCOPY, ELECTROCHEMISTRY, VOLTAGE, ALTERNATING CURRENT, ARTIFICIAL INTELLIGENCE, PAINTS, ELECTROLYTES, ELECTRICAL IMPEDANCE, SEA WATER, ORGANIC COATINGS.
Subject Categories : Coatings, Colorants and Finishes
Distribution Statement : APPROVED FOR PUBLIC RELEASE