A violation behavior recognition method for substation operators based on three-dimensional data and object detection

• Main Authors: Wang, Jing, Zhang, Xiangquan, Zhou, Hualing, Nie, Jianglong, Sun, Han, Chen, Zhao, Su, Zhantao
• Format: Conference Proceeding
• Language: English
• Subjects: Behavioral sciences; Image recognition; Meters; Object detection; personnel positioning spatiotemporal information; Safety management; Solid modeling; Substations; three-dimensional space model; violation behavior recognition
• ISSN: 2693-289X
• DOI: 10.1109/ITOEC57671.2023.10292085

Substation inspection or construction still relies on manual operation. Safety accidents caused by violation behaviors or insufficient knowledge and skills are inevitable. The existing methods for substation operation safety management use two-dimensional map and image to recognize violation behavior including fast moving, cross-border, trespass, high-altitude operation without safety tether, ladder operation without support. However, the accuracy of distance calculation based on two-dimensional map is limited to meter level, which belongs to extensive safety management. Furthermore, it's difficult to capture accurate spatial position information in two-dimensional images, which pose challenges to violation behavior recognition of substation operators. To tackle the above limitations, this paper presents a violation behavior recognition method for substation operators based on three-dimensional data and object detection. Based on three-dimensional space model and personnel positioning spatial-temporal information, it improves the accuracy of safety distance calculation among personnel, equipment and authorized work areas to the decimeter level. The experiments demonstrate that the proposed method can improve the accuracy of fast moving, cross-border, trespass, high-altitude operation without safety tether, ladder operation without support recognition.