Improved multiobjective decision-making model of construction safety supervision in Internet of Things by differential evolution algorithm

Aiming at the problems of slow convergence speed and insufficient solution set diversity in multi-objective optimization of traditional construction safety supervision models, this paper proposes a dynamic weight adaptive differential evolution algorithm (DWA-DE), combined with real-time monitoring data of the Internet of Things, to construct a multi-objective decision-making model for construction safety supervision. The algorithm introduces a local search mechanism based on the characteristics of Internet of Things data by dynamically adjusting the scaling factor and crossover probability, effectively balancing the global exploration and local development capabilities. The experiment uses the MATLAB R2023a platform and uses 3 sets of real construction scene data sets (including 1,200 sensor nodes and 500 hours of monitoring data) for verification. The results show that DWA-DE improves 28.7% in IGD indicators, 41.3% in solution set diversity, and 35.2% in average convergence generations compared with traditional DE. Compared with NSGA-II and MOEA/D, DWA-DE reduces HVR by 19.6% and 24.1% respectively, and the calculation time is shortened to 3.2 seconds/time (traditional DE is 5.8 seconds/time). Through actual verification of a subway project, the model can warn high-risk construction areas 48 hours in advance, with a false alarm rate of only 2.3%, which is significantly better than the existing methods. This study provides an efficient and reliable decision-making tool for IoT construction safety supervision.