Pipeline burst risk early warning method

Abstract

This invention discloses a method for early warning of pipeline burst risks, comprising: collecting multi-source data of the pipeline network and preprocessing it to construct a pipeline network domain knowledge graph containing the pipeline network topology and semantic relationships; reconstructing the hydraulic state of the entire network based on a physical information neural network, inverting the pressure field and velocity field of the pipeline network monitoring blind area under sparse monitoring conditions to obtain dynamic physical field data; extracting the semantic features of the knowledge graph and the temporal features of the dynamic physical field data through a preset burst early warning model, generating high-order risk features through feature fusion, and then calculating the burst risk probability, outputting the burst risk probability of each pipe segment of the pipeline network and triggering corresponding graded early warnings. This invention can output accurate burst risk probabilities and trigger graded response mechanisms in a timely manner, thereby effectively improving the operational safety and emergency management level of the water supply system.

Images