Prediction and early warning method of land subsidence along the railway area based on CNN-LSTM model

Abstract

The invention discloses a method for predicting and early warning of land subsidence in a region along a railway based on a CNN-LSTM model. The CNN-LSTM combined model proposed by the method improves the multi-dimensional feature extraction ability and enhances the prediction accuracy of the model; at the same time, a method based on The grid unit division method for predicting the land subsidence value of the area along the railway line is used to predict the individual subsidence value of each grid unit in the monitoring area; in addition, a method based on the average cumulative subsidence value, average subsidence velocity, and subsidence area is proposed for the ground subsidence along the railway area. , the maximum settlement cumulative value, the settlement non-uniformity coefficient and other settlement prediction index data to carry out settlement risk early warning calculation processing; and proposed three common site tunnels, bridges, subgrade early warning models of railway ground subsidence. For tunnels, bridges, and roadbeds, different settlement prediction index data are selected for risk early warning calculation and processing, and more accurate early warning results are obtained.

Description

technical field [0001] The invention relates to the field of prediction and early warning of geological disasters based on machine learning, in particular to a method for prediction and early warning of ground subsidence in areas along railway lines based on a CNN-LSTM model. Background technique [0002] Land subsidence, also known as ground subsidence or ground subsidence, refers to the regional continuous and slow downward movement of the ground surface under the influence of natural factors and human factors. Land subsidence has always been one of the geological hazards worthy of attention. [0003] With the continuous improvement of the running speed of high-speed trains, in order to ensure the safety and comfort of trains, higher requirements are put forward for the flatness of the ground in the area along the railway. For a high-speed train running at a speed of more than 250 km / h, if there is an uneven track, the vibration effect will be much greater than that of or...

AI Technical Summary

Problems solved by technology

[0009] Aiming at the limitations of the existing subsidence geological disaster risk prediction method in the area along the railway for ground subsidence monitoring and early warning, such as insufficient adaptability and low accuracy, the present invention proposes an artificial neural network based on convolutional neural network (CNN) and long-term short-term memory (LSTM) model for the prediction method of land subsidence value in the area along the railway, and select the relevant subsidence risk monitoring indicators of various railway sites (such as tunnels, bridges, roadbeds) to evaluate the level of railway land subsidence risk, so as to achieve better adaptability and reliability. Accurate early warning results

Images