Research on accident knowledge map establishment and intelligent disposal decision method of power grid construction engineering

The construction of power infrastructure not only involves complex engineering techniques but also faces severe safety challenges. This paper proposes a multi-dimensional information integration and intelligent analysis framework aimed at the field of power infrastructure construction safety, designed to achieve comprehensive and real-time convergence of safety information from the construction site by synthesizing multi-source heterogeneous data. By incorporating the BAGAT model and graph attention networks, the framework is capable of intelligently analyzing the complex conditions of the construction site and accurately identifying potential safety risks and violations. Additionally, this study innovatively applies deep learning-based object detection and semantic recognition technologies to process images from construction sites, automatically extracting key elements from the images and accurately classifying and locating them. The methods and technologies proposed in this paper hold significant application value and practical significance in the field of power infrastructure construction safety, providing robust technical and theoretical support for advancing the intelligent development of this domain.