Neural network methods for corrosion data reduction

✍ Scribed by R.A Cottis; Li Qing; G Owen; S.J Gartland; I.A Helliwell; M Turega

• Book ID: 104314278
• Publisher: Elsevier Science
• Year: 1999
• Weight: 278 KB
• Volume: 20
• Category: Article
• ISSN: 0261-3069
• DOI: 10.1016/s0261-3069(99)00026-6

✦ Synopsis

Neural network methods have been used by a number of workers to model corrosion behaviour. These attempts have generally been regarded as successful, although few have really been able to demonstrate that the neural network model describes the underlying corrosion problem accurately. The application of neural network methods to model the pitting corrosion behaviour of a stainless steel as a function of solution composition and temperature is presented. The predictions of the neural network exhibit reasonable correlation with data for simple one-and two-component solutions, although the problem of testing the generality of the neural network solution remains unsolved. In addition to the basic testing of the neural network performance, there are problems of variability of corrosion behaviour and the unpredictable behaviour of the neural network that may occur in regions of the problem domain where no data are available. The use of simulated data to test the neural network method in conditions similar to those being modelled is suggested as one method of obtaining a better assurance of the applicability and performance of the method.