A slope emergency early warning method, device and equipment for extreme working conditions and a storage medium

Abstract

This application discloses a slope emergency early warning method for extreme working conditions, comprising: collecting long-term multi-source monitoring time-series data of existing slopes, and constructing a source domain slope displacement database after data processing; training a source domain prediction model based on the source domain slope displacement database to learn the spatiotemporal characteristics of slope displacement evolution; for a new target slope, transferring the low-level network parameters in the source domain prediction model to the target prediction model; acquiring short-term monitoring data of the new target slope, and fine-tuning the parameters of the high-level prediction module of the target prediction model based on the short-term monitoring data; performing rolling prediction of the future displacement of the new target slope based on the adapted target prediction model, determining the risk level by combining external rainfall triggering factors, and outputting corresponding early warning information and linkage emergency response instructions. This application can achieve real-time displacement prediction and risk assessment under small sample conditions, improving the timeliness, accuracy, and cross-scenario generalization ability of emergency early warning.

Images