Ultrasonic based concrete defects identification via wavelet packet transform and GA-BP neural network

Area of Science:

• Civil Engineering
• Materials Science
• Non-Destructive Testing

Background:

• Concrete's widespread use in construction necessitates reliable defect detection for structural integrity.
• Current ultrasonic diagnosis methods for concrete defects face challenges due to signal complexity and uncertainty, limiting accuracy and automation.
• Existing techniques struggle to meet the increasing demands for high-performance, automated structural health monitoring.

Purpose of the Study:

• To develop an advanced diagnostic model for identifying concrete defects using ultrasonic information.
• To enhance the accuracy, efficiency, and automation of ultrasonic testing for concrete structures.
• To propose a specific, effective method for recognizing small-sized concrete hole defects.

Main Methods:

• Signal effective information extraction using wavelet packet transform (WPT), employing statistical features like mean, standard deviation, kurtosis, skewness, and energy ratio.
• Defect signal recognition via a genetic algorithm-optimized backpropagation neural network (GA-BPNN) trained on a cross-validated dataset.
• Implementation on 150 ultrasonic detection signals from C30 concrete blocks with varying hole sizes (5, 7, 9 mm) at a 50 kHz working frequency.

Main Results:

• The GA-BPNN model demonstrated a high average recognition accuracy of 91.33% for small concrete defects.
• The proposed signal recognition method effectively characterizes detection signals using WPT-derived features.
• Experimental validation confirmed the feasibility and efficiency of the developed diagnostic model.

Conclusions:

• The GA-BPNN model offers a significant improvement in the accuracy and automation of ultrasonic concrete defect detection.
• The integrated approach of WPT feature extraction and GA-BPNN provides a robust solution for identifying small concrete defects.
• This research contributes to advancing non-destructive testing techniques for concrete structures, enhancing safety and efficiency.