Development of Novel Supervisory System of Underground Confined Space Operation

Confined space operations are prone to cause casualties due to the presence of toxic and harmful substances or lack of oxygen. This paper analyzes the main causes of accidents in confined spaces and provides an evaluation method. Terminal hardware framework and Chip circuit schematic of acquisition unit main control chip is designed. Combining technologies such as improved YOLOX and Internet+, a confined space safety monitoring system architecture is proposed, and an intelligent safety monitoring system for confined space operations covering daily operation management, operation preparation and real-time monitoring is designed. An AI recognition model for unsafe behavior recognition and a deep learning algorithm based on improved YOLOX are proposed. The developed intelligent system helps to effectively reduce the risks of working in confined spaces and improve safety management and control capabilities.