GCN-Based Pollution Source Screening Algorithm and Optimization for Underground Drainage Pipe Network

Source tracing is an important solution to address pollution in underground drainage networks. However, due to the complex and variable hydrological conditions of underground drainage networks, it is still very difficult to accurately trace the pollution sources of underground drainage networks. A graph convolutional neural network can effectively extract structural information from graph data. In this paper, the graph convolutional neural network is used for pollution tracing of underground drainage networks, achieving high accuracy in determining the number of pollution sources and the location of pollution sources.