Power distribution network fault positioning method and system based on physical information neural network

The invention discloses a power distribution network fault positioning method and system based on a physical information neural network. The method comprises the following steps: collecting current values and voltage values of nodes in a power distribution network under different conditions to construct feature data; a power distribution network is divided into a specific number of double-end branch-free line sections, a line where a fault is located is determined based on a current difference value of nodes at the two ends of the line sections, and label data is constructed; preprocessing and normalizing the feature data and the label data, and constructing a model training sample set and a verification sample set; a loss function is constructed by taking minimization of an error between a fault line distance and an actual fault line distance as a target, and a power distribution network fault zero-sequence constraint equation is taken as regularization of a physical constraint to be added into the loss func