Non-contact electroluminescence defect detection method

1 . A non-contact electroluminescence defect detection method, comprising:

irradiating a moving solar cell with a laser to obtain an irradiation area, and simultaneously acquiring a center photoluminescence image, and a left electroluminescence image and a right electroluminescence image based on the irradiation area;

performing a fusion processing on the center photoluminescence image, the left electroluminescence image, and the right electroluminescence image to identify overlapping boundary regions, and conducting a shape analysis on the overlapping boundary regions to determine a boundary distortion, and locating initial defect position coordinates based on the boundary distortion;

measuring a left diffusion distance and a right diffusion distance based on the left electroluminescence image, and the right electroluminescence image, generating a diffusion asymmetry degree based on the left diffusion distance and the right diffusion distance, and determining a defect shift direction based on the asymmetry degree, and correcting the initial defect position coordinates based on the defect shift direction to generate accurate defect position coordinates;

extracting a center luminescence intensity from the center photoluminescence image, and extracting an edge luminescence intensity from the left electroluminescence image, and the right electroluminescence image, and forming a luminescence gradient curve based on the center luminescence intensity and the edge luminescence intensity, and identifying intensity jumping points through the luminescence gradient curve;

adopting the intensity jumping points as dividing points to segment a luminescence response interval, establishing an energy distribution index based on the luminescence response interval, identifying energy conversion defects at the accurate defect position coordinates based on the energy distribution index;

conducting a conduction path analysis based on the accurate defect position coordinates and the energy conversion defects to determine a defect coupling coefficient; identifying defect types based on the defect coupling coefficient, and outputting a spatial positioning map based on the defect types, completing the non-contact electroluminescence defect detection.

2 . The non-contact electroluminescence defect detection method according to claim 1 , wherein acquiring center photoluminescence image and left electroluminescence image, and the right electroluminescence image based on the irradiation area comprises:

establishing a light signal timestamp mapping relationship within the irradiation area;

collecting complementary signals of center photoluminescence single, left electroluminescence signal, and right electroluminescence signal based on the timestamp mapping relationship to generate complementary signal pairs;

performing spatiotemporal cross validation on the complementary signal pairs to form a verification matrix;

forming the center photoluminescence image based on the verification matrix, and the left electroluminescence image, and the right electroluminescence image.

3 . The non-contact electroluminescence defect detection method according claim 1 , wherein performing a fusion processing on the center photoluminescence image, the left electroluminescence image, and the right electroluminescence image to identify overlapping boundary regions comprises:

extracting boundary cross-section shapes from the center photoluminescence image, the left electroluminescence image, and the right electroluminescence image;

performing a multi-angle projection analysis based on the boundary cross-section shapes to generate projection contour sets;

performing a cross-overlay processing on the projection contour sets to form an overlay map;

identifying boundary overlap regions based on the overlay map.

4 . The non-contact electroluminescence defect detection method according to claim 1 , wherein generating a diffusion asymmetry degree based on the left diffusion distance and the right diffusion distance comprises:

forming diffusion trajectories based on the left diffusion distance and the right diffusion distance;

identifying diffusion acceleration segments and diffusion deceleration segments from the diffusion trajectories;

conducting a duration comparison analysis on the diffusion acceleration segments and the diffusion deceleration segments to generate a duration difference value;

generating the diffusion asymmetry degree based on the duration difference value.

5 . The non-contact electroluminescence defect detection method according to claim 1 , wherein forming a luminescence gradient curve based on the center luminescence intensity and the edge luminescence intensity comprises:

establishing an intensity difference sequence based on the center luminescence intensity and the edge luminescence intensity;

performing a gradient stratification on the intensity difference sequence to generate steep gradient layers, gradual gradient layers, and gentle gradient layers;

extracting mutation features from the steep gradient layers, and extracting transition features from the gradual gradient layers, and extracting steady-state features from the gentle gradient layers;

modulating and reconstructing the mutation features, the transition features, and the steady-state features to generate the luminescence gradient curve.

6 . The non-contact electroluminescence defect detection method according to claim 1 , wherein establishing an energy distribution index based on the luminescence response interval comprises:

identifying energy-dense segments and energy-sparse segments within the luminescence response interval;

calculating concentration indices based on the energy-dense segments, and calculating dispersion indices based on the energy-sparse segments;

performing an energy balance evaluation based on the concentration indices and the dispersion indices to generate balance parameters;

generating the energy distribution index based on the balance parameters.

7 . The non-contact electroluminescence defect detection method according to claim 1 , wherein conducting a conduction path analysis based on the accurate defect position coordinates and the energy conversion defects to determine a defect coupling coefficient comprises:

establishing defect impact boundaries based on the accurate defect position coordinates;

tracking a conduction path of the energy conversion defects within the defect influence boundaries, to identify a conduction enhancement path and a conduction attenuation path;

determining the defect coupling coefficient based on a length ratio of the conduction enhancement path and the conduction attenuation path.

8 . The non-contact electroluminescence defect detection method according to claim 3 , wherein performing a multi-angle projection analysis based on the boundary cross-section shapes to generate a projection contour sets comprises:

evaluating a projection complexity of the boundary cross-section shapes to determine a projection strategy, the projection complexity comprises a boundary curvature, a defect distribution density, and a boundary continuity;

setting a multi-angle projection scanning sequence based on the projection strategy;

performing a profile extraction for each of angles in the multi-angle projection scanning sequence to generate a projection contour sets.

9 . The non-contact electroluminescence defect detection method according to claim 5 , wherein modulating and reconstructing the mutation features, the transition features, and the steady-state features to generate the luminescence gradient curve comprises:

conducting polarity identification on the mutation features, the transition features, and the steady-state features to generate a positive polarity feature group and a negative polarity feature group;

neutralizing the negative polarity feature group based on the positive polarity feature group to form balanced features;

modulating and encoding the balanced features to generate a feature modulation signal;

demodulating and reconstructing the feature modulation signal to generate the luminescence gradient curve.