Construction Safety Decision-Making System Based on Improved Integration of Case-Based Reasoning and Rule-Based Reasoning

With the rapid development of the construction industry, safety management at construction sites faces increasingly complex challenges, particularly due to inaccurate decision-making processes that may lead to safety incidents. Existing decision-making systems, although widely used, often have inefficiencies, high costs, and limited adaptability, making them inadequate for complex construction environments. This paper presents a construction safety decision support system integrating case-based reasoning (CBR) and rule-based reasoning (RBR), aiming to enhance the timeliness and scientific accuracy of decisions through intelligent technologies. The proposed system leverages historical case data for experience-based support and standardizes the decision-making process through rule-based reasoning, significantly improving decision response speed and accuracy. It incorporates a transformer-based self-attention mechanism to optimize the CBR retrieval process for more accurate and rapid case matching. The system also enhances the K-nearest neighbors (K-NN) algorithm with convolutional neural networks (CNNs) to dynamically adjust case attribute weights, reflecting decision requirements more precisely. Additionally, a genetic algorithm and reinforcement learning-based model is employed to intelligently revise historical case content and rules to align with target scenarios applicable decision plans. Validation of the proposed system was conducted using historical data, demonstrating significant improvements in decision-making time and accuracy. A case study of a high-rise building construction project showed an increase in safety hazard identification rates from 70% to 83% and a 10% improvement in decision accuracy, confirming the system's effectiveness and potential value in complex construction safety management. This research provides valuable insights and practical solutions for enhancing construction safety management in complex environments.