How Intelligent Data Mining Techniques Improve Decision-Making and Reduce Uncertainty in Structural Engineering Research

This systematic review of studies investigates the extent to which intelligent data mining techniques can reduce uncertainty and enhance decision-making in structural engineering. Findings reveal strong but highly context-dependent effectiveness: reported benefits range from % cost reduction in seismic dam design (Amini et al.) to % damage-detection accuracy under noisy and delayed sensor signals (Salehi et al.). Pattern-recognition methods (ANNs, SVMs, decision trees, PPSVM) proved particularly effective for measurement-uncertainty problems in damage detection and structural health monitoring, whereas surrogate modelling, decision-tree-based probabilistic approaches (DTUD), and design-space-reduction techniques showed clear value for parameter and model uncertainty in computationally expensive structural optimization tasks. Data-fusion and evidence-theory integrations further improved robustness against conflicting sensor information. However, rigorous quantitative validation remains scarce: only two studies provided specific numerical performance metrics, none reported statistical significance, confidence intervals, or uncertainty bounds on results, and nine sources were available only as abstracts, severely limiting assessment of methodological quality. Effectiveness therefore depends critically on problem type, data quality, scale (laboratory vs. full-scale), and proper technique selection, with generalizability across structural engineering applications still poorly characterized and requiring substantially more rigorous, full-text empirical validation.

Keywords:

Intelligent data mining , Machine learning , Structural engineering , Structural optimization