Particle Swarm Optimization approach to defect detection in armour ceramics

Area of Science:

• Materials Science
• Non-destructive Testing
• Artificial Intelligence

Background:

• Automated inspection systems are crucial for quality control in ceramic component manufacturing.
• High dimensionality and redundant features in datasets can significantly impair classification performance.
• Evolutionary algorithms are effective for multi-criteria optimization problems like feature selection.

Purpose of the Study:

• To develop an automated ultrasonic sensor-based system for defect classification in ceramic components.
• To investigate the efficacy of Binary Coded Particle Swarm Optimization (BPSO) for feature subset selection in this context.
• To optimize classification error rates by reducing dataset dimensionality.

Main Methods:

• Development of an automated ultrasonic inspection system.
• Application of Binary Coded Particle Swarm Optimization (BPSO) for feature subset selection.
• Utilizing an Artificial Neural Network (ANN) as a fitness function evaluator for BPSO.

Main Results:

• BPSO effectively reduced the number of features while optimizing classification error rates.
• The proposed BPSO-based feature selection improved the performance of the defect classification system.
• Demonstrated the potential of BPSO in classifying high-frequency ultrasonic signals for defect detection.

Conclusions:

• The integration of BPSO with an ANN provides an effective approach for feature selection in automated ultrasonic inspection.
• This method enhances the accuracy and efficiency of defect classification in ceramic components.
• The study highlights BPSO as a promising technique for optimizing high-frequency ultrasonic signal classification.