School of Automation Engineering, Northeast Dianli University, Jilin, China

---

Abstract—In this paper, the level of corrosion and the corrosion rate of 304 stainless steel induced by sulfatereducing bacteria were studied using electrochemical noise. The noise data were analyzed by time domain and frequency domain combined with the observations of optical microscope. And the corrosion was divided into four categories: passivation, pitting induction period, pitting and uniform corrosion. The traditional method for electrochemical noise analysis has lag shortcomings, so the feasibility study on Hilbert-huang Transform and BP Neural Network on intelligent recognition method for microbiologically influenced corrosion was conducted. The results showed that the use of Hilbert-huang Transform for feature extraction can characterize the level of corrosion;BP Neural Network could identify passivation, pitting induction period and pitting correctly, and recognition effect for uniform corrosion would be improved. A feasible way of analyzing electrochemical noise data real-time and intelligent was provided on this paper, and it was hoped that the analyzing method could provide theoretical basis in the identification of the extent of corrosion in practice to take preventive measures timely.

Index Terms—Microbiologically influenced corrosion, Hilbert-huang Transform, BP Neural Network, identification.

[PDF]

Cite: Hong Men, Jing Zhang, Lihua Zhang, "Intelligent Recognition for Microbiologically Influenced Corrosion Based On Hilbert-huang Transform and BP Neural Network," Journal of Computers vol. 7, no. 9, pp. 2283-2291, 2012.