Integrating Fuzzy Logic and Deep Neural Networks for Intelligent Construction Quality Management in Industrial Informatics

This research drew upon 987 construction inspection data points spanning from 1993 to 2023, sourced from the Taiwanese Public Construction Management Information System, to unveil the correlation between construction elements and quality outcomes. Initially, fuzzy logic was employed to compute the weights of 499 defects, resulting in the identification of 25 pivotal construction factors based on these weight allocations. Subsequently, a deep neural network was deployed to discern the interrelation between these significant construction factors (input variables) and the resultant construction quality (output variable). The evaluation of the prediction model's performance substantiated the influence of these key construction factors on project outcomes. In line with the contemporary trend of intelligent industrial informatics, this study harnessed the application of machine learning to enable these systems to adapt and learn from historical data. The developed hybrid soft computing approach, amalgamating fuzzy logic and artificial neural networks, demonstrated an impressive accuracy rate of 95.95%. The insights gleaned from this research offer project managers actionable intelligence to enhance project management efficacy and establish robust construction management protocols, thereby elevating the overall construction quality of projects.