Evaluation on Quantitative Recognition of Micro Cracks by Magnetic Leakage Test Based on GA-BP Neural Network

The basic principle of realizing quantitative evaluation of metal micro crack detection with using BP neural network optimized by genetic algorithm is introduced. Organic combination of genetic algorithm and artificial neural network not only improves global search performance, but also maintains good adaptability to nonlinear problems during magnetic flux leakage detection. Final experimental results show that the artificial intelligence algorithm applied in practical engineering can realize quantitative assessment of metal micro cracks based on magnetic leakage signals.