Quantitative detection of rebar corrosion by magnetic memory based on first-principles

Abstract

Abstract Reinforcement corrosion seriously impacts the bearing capacity and durability of reinforced concrete(RC) structures. It is very important to detect reinforcement’s corrosion state in concrete timely and effective. This paper introduced the magnetic memory method to the quantitative detection of reinforcement corrosion. Based onfirst- principles, the causes of material magnetism were explained from the point of view of particles. The microscopic models of reinforcement corrosion were established and the correlation between the absolute value of magnetization M and mass loss rate α had been analyzed. The experiment of magnetic memory testing of the rebar corrosion was carried out, and the magnetic induction curves of the rebars at different mass loss rates were obtained. Finally, the random forest algorithm was used to realize the quantitative recognition of steel corrosion. The results of microscopic models showed that |M| increased nonlinearly with α. The tangential and normal magnetic induction curves obtained by the experiment showed a trend of overall movement and increasing volatility with the increase of α, then four magnetic indexes (I 1xn , I 1zn , I 2xn , I 2zn ) were proposed based on tangential and normal magnetic induction curves to characterize the mass loss rate α. The regularity of I-α curves was consistent with that of |M|-α curves obtained by the microscopic model. The random forest algorithm was introduced to solve the nonlinear and discrete problems of magnetic indexes, and a hierarchical prediction model of rebar corrosion was established. The prediction accuracy of the model was 85.7%, which can realize the state recognition of steel bars under low corrosion degrees.