Grading assessment of steel bar corrosion in concrete structures based on naive Bayesian model and spontaneous magnetic flux leakage effect

Abstract

In order to explore an accurate method to evaluate the corrosion of reinforced concrete structures, the spontaneous magnetic flux leakage (SMFL) signal distribution on the surface of reinforced concrete specimens under different corrosion degrees was scanned based on SMFL technology. The influence of steel bar length, steel bar diameter, and other parameters on the distribution of SMFL signal was studied. The correlation between steel bar corrosion and the characteristic magnetic index of concrete structure was explored. Based on the naive Bayesian model, the classification evaluation of the steel bar corrosion degree of concrete structure was carried out. The results show that the variation of SMFL signal is affected by the corrosion degree α. When the lift-off height and the thickness of concrete protective layer remain unchanged, the slope between the peak and trough of Bz (magnetic induction intensity along z direction) curve increases with the increase of α, and the trough of Bx (magnetic induction intensity along x direction) curve decreases with the increase of the corrosion degree α. The peak and trough of magnetic signal curve can be used as the basis for determining the corrosion position. There is a strong correlation between the magnetic characteristic index β, γ, and the steel corrosion degree α obtained by SMFL. Through the characterization relationship between α, β, and γ, the corresponding models of single and comprehensive index β and γ were established. The results showed that the accuracy of β and γ integrated discriminant Naive Bayesian model-III reached 90.7%, which proved that the evaluation method has high reliability. This study explores the application of SMFL in corrosion detection of concrete structures.