Magneto-optical imaging deviation model of micro-gap weld joint

Seam tracking is important for butt joint laser welding. A magneto-optical imaging approach is proposed to detect the micro-gap weld whose width is less than 0.2 mm. The symmetry of the magnetic field above the weld joint is an important precondition to ensure the detection accuracy of the magneto-optical imaging method. However, in actual complex industrial scene, it is difficult to guarantee complete symmetry of the magnetic field. This paper proposes an effective method for solving the problem of magneto-optical imaging deviation under an asymmetric magnetic field. Two possible factors causing the asymmetry of magnetic field above weld joint are firstly investigated using finite element analysis. By analyzing the characteristics of the magneto-optic medium used in the sensor and measuring the magnetic field distribution of weld joint at different lift-off height and different excitation voltage, the prediction model of deviation between the weld position detected by magneto-optical imaging and the actual weld position is built by back propagation (BP) neural network. The experimental result of weld seam tracking based on magneto-optical imaging shows that the change of the lift-off height will affect the detection accuracy of the weld position, and sufficient accuracy can be ensured after correcting the deviation according to the prediction model of magneto-optical imaging deviation.