Hybrid model-based digital twin modeling method and system for power communication network

The invention discloses an electric power communication network digital twin modeling method and system based on a hybrid model, and the method comprises the steps: obtaining multi-station and multi-level three-dimensional point cloud data based on a high-precision three-dimensional laser scanner, and employing ResPointNet + + for scanning the semantic segmentation of a point cloud, so as to recognize an electric power communication network resource component; in the semantic segmentation process, decoding is carried out through an improved depth kernel extreme learning machine algorithm to recover spatial information; establishing a mechanism model which objectively and accurately reflects the dynamic characteristics of the power communication network and the power grid coupling network, and obtaining an operation mechanism between each component in a physical model; and constructing a data driving model, and driving a physical model to reversely control a physical network layer by utilizing the characterist