The application discloses an
environmental safety abnormal event discovery and tracing method and
system based on a space-time graph network, belongs to the field of
environmental safety monitoring, and aims at solving the problems that abnormal events are difficult to be stably positioned, it is difficult to distinguish the contribution of multiple sources, there is a lack of auditable transmission evidence, and it is easy to be disturbed by sensor faults, because the influence relationship between monitoring stations dynamically changes with external conditions such as
wind direction,
wind speed, flow direction, flow rate
and gate pump working conditions. The application forms a dynamic graph by constructing an initial
directed graph, calculating edge flux prior, combining residual correction and fusion to obtain corrected edge flux, predicts and detects abnormalities based on a space-time graph network including
convection update and
diffusion update, and further searches for a transmission path and extracts a key time window based on
time integral of edge contribution degree to calculate a tracing confidence to generate an evidence chain, so that the technical effects of improving the accuracy of abnormal detection and tracing, sorting source contribution degree and making the tracing result interpretable and reproducible are achieved.