From Cognitive Bias Toward Advanced Computational Intelligence for Smart Infrastructure Monitoring

Area of Science:

• Engineering
• Computer Science
• Artificial Intelligence

Background:

• Visual inspections for infrastructure condition assessment are subjective and prone to cognitive bias.
• Cognitive biases, such as confusing condition state with safety, impact Structural Health Monitoring (SHM) decision-making.
• Computer-based approaches offer objective and reliable solutions for SHM systems.

Purpose of the Study:

• To explore the integration of advanced computational intelligence with SHM solutions.
• To develop and evaluate AI-based algorithms for infrastructure damage assessment.
• To mitigate the impact of human judgment errors in SHM.

Main Methods:

• Development of Artificial Neural Network (ANN) and hybrid ANN algorithms (Imperial Competitive Algorithm, Genetic Algorithm).
• Application of these algorithms for damage assessment on a lab-scale composite bridge deck structure.
• Comparative analysis of algorithm performance to evaluate improvements in prediction accuracy.

Main Results:

• Evolutionary algorithms (hybrid ANN-based ICA and GA) significantly improved the prediction error of the pre-developed ANN.
• Enhanced learning procedures in ANN led to more accurate damage assessment.
• Demonstrated the efficacy of AI in overcoming limitations of traditional visual inspections.

Conclusions:

• Advanced AI algorithms, particularly evolutionary approaches, offer a robust enhancement for SHM systems.
• Computational intelligence provides a powerful tool for objective and accurate infrastructure condition assessment.
• The developed methodology shows promise for reliable and unbiased structural health monitoring.