A subway tunnel water seepage risk early warning method based on interval video stream

Abstract

The application discloses a subway tunnel water seepage risk early warning method based on interval video stream and concretely relates to the field of industrial data processing and acquisition, and is used for solving the problems of water seepage false alarm and benchmark pollution caused by transient optical disturbance in subway high-speed dynamic video stream; the global picture brightness gradient variation integral parameter is calculated for the video frame sequence of a target monitoring interval, the suspended interception is executed, the released video frame sequence is input into a decoupling encoder to complete hidden space mapping, the steady-state structure hidden vector sequence representing tunnel wall texture is output, then the long-acting background update window and the short-acting observation window are formed based on double-scale time windows, the data stream in the long-acting background update window is used to determine the topological ball center benchmark coordinate of the updated single-class space measure model, and finally, when the drift displacement distance of the steady-state structure hidden vector sequence accumulated in the short-acting observation window relative to the topological ball center benchmark coordinate presents a continuous expansion attribute and exceeds the peripheral safety boundary, a water seepage prevention instruction is output.

Images