Efficient rebar dimension detection using prior-guided discrete global optimization full waveform inversion

Abstract. In non-destructive rebar dimension detection in concrete using full waveform inversion (FWI), exhaustive search methods and global optimization algorithms struggle with nonlinear electromagnetic responses and computational complexity, hindering their integration into engineering detection systems. This paper proposes a prior-guided discrete global optimization full waveform inversion (PG-DGOFWI) method to improve the efficiency and accuracy of rebar dimension detection. The method first uses reliable prior information to convert rebar coordinates and dimensions from continuous to discrete arrays, reducing the search space. It then constructs an objective function from electromagnetic wave data and identifies parameters using a global optimization algorithm. Simulation using the gprMax compared the inversion results of exhaustive search, global optimization, and PG-DGOFWI methods. Experimental tests based on the simulation model showed that PG-DGOFWI significantly improves rebar dimension detection. This method uses prior information to guide the search, reducing computational complexity and avoiding local optima, thereby enhancing stability and reliability.