In an actual structure, the arrangement of steel bars is complicated, there are many factors affecting the corrosion of steel bars, and these factors affect each other. However, accurately reflecting the corrosion of steel bars in actual engineering through theoretical calculations is difficult. Besides, it is impossible to detect and evaluate steel bars rust completely and accurately. This article is based on spontaneous magnetic leakage detection technology and adopts the method of stage corrosion and scanning along the reinforcing bar. Based on spontaneous magnetic flux leakage detection technology, the linear change rate of the tangential component curve of the magnetic flux leakage signal generated after the corrosion of a steel bar is studied, and a comparison is made between the steel bar coated concrete samples with different steel bar diameters. In this paper, the “origin of magnetic flux leakage signal” is defined as a reference point, which is convenient for effectively comparing the magnetic signal curves under all operating conditions. Besides, the “rust-magnetic fluctuation parameter” is proposed to accurately reflect the sudden change of leakage magnetic field caused by disconnection due to the corrosion of a steel bar. A new data processing method is provided for the non-destructive testing of steel corrosion using the spontaneous magnetic flux leakage effect, which can effectively reduce the influence of steel bar diameter on magnetic flux leakage signal and improve the precision of non-destructive testing technology of steel bar corrosion using the metal magnetic memory effect.

In order to solve the problem of quantitative detection of corroded reinforced concrete of different sizes, the quantitative detection experiment based on spontaneous magnetic flux leakage (SMFL) was carried out in batches. Electrochemical corrosion of 27 reinforced concrete specimens was carried out, and the SMFL signals of reinforcement were obtained by magnetic detection equipment. Four-dimensional magnetic indicators M₁–M₄ that can characterize the corrosion degree of the specimens were defined. The influence of different sizes on the magnetic indicators were analyzed. It is concluded that the thicknesses of the concrete covers affect the magnetic indicators by affecting the lift-off heights z. The influence of diameter of the rebars on the quantitative detection can be eliminated by describing the corrosion degree with the average cross-section loss rate α. The influence of length of the rebars on quantitative detection is not clear yet. Finally, Support Vector Machine (SVM) was introduced to establish a classification model of corrosion classes and magnetic indicators. Using the model to predict the corrosion classes of the specimens can achieve a high classification accuracy. The research provides a new method for the quantitative detection of steel corrosion.