Defects Prediction Method for Radiographic Images Based on Random PSO Using Regional Fluctuation Sensitivity.

Area of Science:

• Radiographic imaging analysis
• Computational intelligence
• Materials science

Background:

• Conventional particle swarm optimization (PSO) struggles with precise defect localization in radiographic images due to premature convergence and a non-defect-centric approach.
• Stable velocity PSO models often fail to accurately identify defect regions, impacting the reliability of defect prediction systems.

Purpose of the Study:

• To develop an advanced defect prediction methodology for radiographic images using a refined particle swarm optimization (PSO) algorithm.
• To enhance the precision and efficiency of defect detection by introducing a fluctuation-sensitive approach (FS-PSO).

Main Methods:

• Development of a fluctuation-sensitive particle swarm optimization (FS-PSO) model.
• Modulation of movement intensity based on swarm size to reduce chaotic movement and enhance efficiency.
• Rigorous evaluation through simulations and practical blade experiments.

Main Results:

• The FS-PSO model demonstrated an approximate 40% increase in particle entrapment within defect areas.
• FS-PSO achieved an expedited convergence rate with a maximal additional time consumption of only 2.28%.
• Empirical findings showed substantial performance improvement over conventional stable velocity models, especially in shape retention for defect extraction.

Conclusions:

• The proposed FS-PSO model offers a significant advancement in defect prediction for radiographic images.
• FS-PSO effectively addresses the limitations of conventional PSO, providing higher accuracy and efficiency in defect localization.
• The enhanced model shows promise for improving non-destructive testing and quality control in relevant industries.