Study On Corrosion Prediction Based On CO₂Corrosive Morphology Methods

CO₂corrosion images and corrosion data are very import and fundamental to diagnosecorrosion type，evaluate corrosion degree and study corrosion mechanism in CO₂corrosivemorphology methods. Manual identification and detection of these images is a tedious work.It is also easy to be affected by subjective factors. Thus, quantitatively analyzing corrosionmorphology, extracting and transforming the corrosion information from numerous corrosionimages into recognizable information by computers obtain the useful knowledge.The carbon dioxide corrosion morphology features obtained from N80steel are extractedusing three typical methods including grey level data matrix statistic, wavelet transform andimage binaryzation process. Pits area has been obtained accurately using binary imageextraction algorithm based on numbers of pixels. The pits numbers are also received by classof pixels. The complex corrosion morphologies are characterized by eigenvalue of energygrey data matrix statistics. Considering the multiplayer feed forward neural networks, apitting velocity diagnosis model has been developed based on two corrosion type criterions,which including the anisotropic energy parameter of corrosion images and the image energyparameter after wavelet transform. The diagnosis results agree well with the testing results,which can show a new way for carbon dioxide corrosion prediction.