Concrete crack detection using ridgelet neural network optimized by advanced human evolutionary optimization

Area of Science:

• Civil Engineering
• Computer Science
• Artificial Intelligence

Background:

• Concrete structures are prone to cracking, which compromises their integrity and durability.
• Accurate crack diagnosis is crucial for maintaining structural safety and performance.

Purpose of the Study:

• To develop an innovative algorithm for concrete crack diagnosis.
• To optimize a Deep Neural Network (DNN) using a novel evolutionary algorithm.

Main Methods:

• An optimized Deep Neural Network (DNN) called the Ridgelet Neural Network (RNN) was developed.
• The RNN model was enhanced using a new Advanced Human Evolutionary Optimization (AHEO) algorithm.
• A labeled image dataset was used for training, with the AHEO algorithm refining weights, adjusting the output layer, and augmenting the dataset.

Main Results:

• The RNN/AHEO model achieved high accuracy (99.665%) and F1-score (99.035%) in concrete crack detection.
• The model demonstrated superior performance compared to existing methods like CNN, CrackUnet, R-CNN, DCNN, and U-Net.
• The AHEO algorithm effectively optimized the RNN for binary classification of cracks.

Conclusions:

• The developed RNN/AHEO model offers a robust and highly accurate solution for concrete crack diagnosis.
• This innovative approach significantly improves the detection of structural weaknesses in concrete frameworks.
• The study highlights the potential of combining advanced neural networks with evolutionary optimization for infrastructure monitoring.