Safety Pattern Recognition Based on Elman Neural Network Method in Tower Crane Applications

Tower cranes, as specialized construction machinery, are widely used in urban construction projects due to their capacity for lifting heavy loads and facilitating high-rise operations, yet they are also associated with a relatively high incidence of operational accidents. Ensuring the safety of tower crane operations is therefore critical for both worker protection and construction project efficiency. This paper proposes the application of an Elman neural network to establish a precise mapping between key state parameters of tower cranes-such as load weight, boom angle, wind speed, and operational posture-and distinct categories of construction safety states. By integrating real-time monitoring data, the model is capable of accurately recognizing and classifying the current safety state of tower cranes under various operational conditions. The proposed method was validated through empirical analysis on actual construction sites, demonstrating that it can effectively identify potential safety hazards, provide early warning signals, and assist in proactive safety management. The results indicate that the Elman neural network approach not only improves the accuracy of safety state recognition but also offers practical guidance for the development of intelligent monitoring systems, contributing to enhanced operational safety and risk mitigation in tower crane applications.