MFL inspection defect reconstruction based on self-learning PSO

As an efficient optimization method, iterative approach plays an important role in the signal inversion of magnetic flux leakage (MFL) technology. Particle swarm optimization (PSO), a new population-based iterative optimization technique, has been applied for many real world problems with promising results. Self-learning particle swarm optimization (SLPSO), a recently proposed variant of PSO, has been proved to have superior performance in diverse global optimization benchmark problems with 100 dimensions or even more. In this paper, as an iterative approach, SLPSO is applied to defect profile reconstruction for magnetic flux leakage inspection. RBFNN is also used as forward model in the SLPSO-based defect reconstruction method. The experimental results show the profiles processed by the SLPSO-based defect reconstruction method are significantly precise.