CN-121557951-B - Surface subsidence monitoring method, device, equipment, storage medium and program product

CN121557951BCN 121557951 BCN121557951 BCN 121557951BCN-121557951-B

Abstract

The application discloses a method, a device, equipment, a storage medium and a program product for monitoring surface subsidence, and relates to the technical field of surface subsidence monitoring. Determining data acquisition interval time of two adjacent InSAR settlement data in the time sequence InSAR settlement data set, performing time coding on the data acquisition interval time to obtain a time interval matrix, and training a ground surface settlement prediction model by adopting the time interval matrix and the time sequence InSAR settlement data set to obtain a target model, wherein the target model is used for monitoring ground surface settlement of a target area to be monitored. The time interval matrix and the time sequence InSAR subsidence data set obtained by the time coding are utilized to train the ground surface subsidence prediction model, so that accuracy errors caused by non-uniform time sequence data are reduced, the adaptability of the target model to irregular sampling data is improved, the prediction accuracy of the target model is further improved, and finally ground surface subsidence is accurately monitored.

Inventors

LI YA
WEI DAOQUAN
GUO LIYE
XUE SHASHA

Assignees

陕西北斗环境信息产业有限公司

Dates

Publication Date: 20260508
Application Date: 20251209

Claims (10)

1. The earth surface subsidence monitoring method is characterized by comprising the following steps: acquiring a time sequence InSAR sedimentation data set, and determining the data acquisition interval time of two adjacent InSAR sedimentation data in the time sequence InSAR sedimentation data set; The time coding is carried out on the data acquisition interval time to obtain a time interval matrix, and the following formula is adopted Time encoding the data acquisition interval, wherein, , As a matrix of weights that can be learned, The representative time of day is indicated by the time, GeLU is a Gaussian error linear unit activation function for the bias value; Building a ground surface subsidence prediction model, and training the ground surface subsidence prediction model by adopting the time interval matrix and the time sequence InSAR subsidence data set to obtain a target model, wherein the target model is used for monitoring ground surface subsidence of a target area to be monitored.
2. The method of monitoring earth's surface subsidence as set forth in claim 1, wherein the step of training the earth's surface subsidence prediction model using the time interval matrix and the time series InSAR subsidence dataset comprises: Determining the time interval matrix and the time attention and the space attention of fusion data corresponding to the time sequence InSAR sedimentation data set; fusing the time attention and the space attention to obtain space-time joint attention; training the surface subsidence prediction model based on the spatiotemporal joint attention.
3. The method of monitoring surface subsidence as set forth in claim 1, wherein the step of constructing a prediction model of surface subsidence comprises: And constructing a ground surface subsidence prediction model by adopting a two-way space-time subsidence characteristic extraction network combined with a convolutional neural network and a long-term memory neural network.
4. The method of monitoring earth's surface subsidence as recited in claim 1, wherein the step of training the earth's surface subsidence prediction model comprises: constructing a sedimentation space-time statistical weighting loss; And training the ground surface subsidence prediction model through the subsidence space-time statistical weighting loss.
5. The method of monitoring surface subsidence as set forth in claim 4, wherein the step of constructing a statistical weighted loss of subsidence time and space comprises: Separately constructing a space-time attention loss and a statistical loss, wherein the statistical loss comprises a correlation loss, a variance loss and a mean loss; And obtaining balance coefficients of the space-time attention loss and the statistical loss, and constructing and obtaining sedimentation space-time statistical weighting loss based on the space-time attention loss, the statistical loss and the balance coefficients.
6. The method of monitoring earth subsidence as set forth in claim 1, wherein the step of training the earth subsidence prediction model using the time interval matrix and the time series InSAR subsidence dataset comprises, after the step of obtaining a target model: Determining a target area to be monitored; and monitoring the surface subsidence of the target area to be monitored through the target model.
7. An earth's surface subsidence monitoring device, characterized in that, the earth's surface subsidence monitoring device includes: the preparation module is used for acquiring a time sequence InSAR sedimentation data set and determining the data acquisition interval time of two adjacent InSAR sedimentation data in the time sequence InSAR sedimentation data set; The coding module is used for carrying out time coding on the data acquisition interval time to obtain a time interval matrix, and the following formula is adopted Time encoding the data acquisition interval, wherein, , As a matrix of weights that can be learned, The representative time of day is indicated by the time, GeLU is a Gaussian error linear unit activation function for the bias value; The application module is used for constructing a ground subsidence prediction model, training the ground subsidence prediction model by adopting the time interval matrix and the time sequence InSAR subsidence data set to obtain a target model, wherein the target model is used for monitoring ground subsidence of a target area to be monitored.
8. A surface subsidence monitoring apparatus comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program configured to implement the steps of the surface subsidence monitoring method of any one of claims 1 to 6.
9. A storage medium, characterized in that the storage medium is a computer-readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, carries out the steps of the surface subsidence monitoring method according to any one of claims 1 to 6.
10. A computer program product, characterized in that the computer program product comprises a computer program which, when executed by a processor, implements the steps of the surface subsidence monitoring method according to any one of claims 1 to 6.

Description

Surface subsidence monitoring method, device, equipment, storage medium and program product Technical Field The application relates to the technical field of earth surface subsidence monitoring, in particular to an earth surface subsidence monitoring method, an earth surface subsidence monitoring device, earth surface subsidence monitoring equipment, a storage medium and a computer program product. Background The traditional ground subsidence monitoring mode mainly utilizes instrument equipment such as total stations, high-precision level gauges, global navigation system (GNSS) and the like to establish monitoring stations on the plane of a deformation area, measures subsidence values and subsidence rates in different time periods, feeds back ground subsidence data in real time, and pre-warns abnormal values exceeding a threshold value. However, the method mainly lays monitoring points in a point-line mode, only elevation change data of discrete monitoring points can be obtained, and deformation trend and space distribution rule of the whole area to be monitored cannot be reflected. In this regard, the development of the synthetic aperture radar interferometry (InSAR) brings new directions to the monitoring of earth surface subsidence, and the technology has the characteristics of being not limited by environmental conditions and being capable of realizing earth observation all the time, all weather, large range and high precision, and no ground control points or contact monitoring areas are required to be arranged. In recent years, with the development of artificial intelligence, a deep learning (DEEP LEARNING) based method is used for amplifying the wonderful colors in the remote sensing field, the technology has strong learning capacity and can solve the problem of complicated nonlinear deformation prediction, the technology does not need to rely on manually defined physical parameters (such as lithology, hydrologic characteristics and the like), the sedimentation characteristics in the time sequence InSAR data can be automatically extracted through a neural network, and the dependence of a traditional physical model on parameter acquisition is reduced. The method solves the bottleneck of the traditional remote sensing method in complex scene interpretation, large-scale data processing and dynamic monitoring through ‌ automatic feature extraction ‌, ‌ multi-source data fusion ‌ and ‌ efficient computing architecture ‌. However, in the process of processing InSAR data, when the remote sensing image is affected by atmospheric interference (such as cloud layer, ionosphere disturbance and the like) or noise, interference patterns are incoherent, and the low-quality data can be automatically removed. In addition, snow coverage or seasonal variations in vegetation may also reduce radar signal coherence. To balance accuracy with data availability, it is necessary to discard portions of the point-in-time data. Due to the fact that different degrees of deletion exist at time points, sedimentation data with uneven periods, namely InSAR data, cannot truly reflect continuous change trend of sedimentation, so that sedimentation prediction accuracy is low, and finally, surface sedimentation monitoring is inaccurate. The foregoing is provided merely for the purpose of facilitating understanding of the technical solutions of the present application and is not intended to represent an admission that the foregoing is prior art. Disclosure of Invention The application mainly aims to provide a ground surface subsidence monitoring method, a ground surface subsidence monitoring device, ground surface subsidence monitoring equipment, a storage medium and a computer program product, and aims to solve the technical problems that due to the fact that missing InSAR data cannot truly reflect continuous change trend of subsidence, the accuracy of subsidence prediction is not high, and finally the ground surface subsidence monitoring is not accurate enough. In order to achieve the above object, the present application provides a method for monitoring surface subsidence, which comprises: acquiring a time sequence InSAR sedimentation data set, and determining the data acquisition interval time of two adjacent InSAR sedimentation data in the time sequence InSAR sedimentation data set; Performing time coding on the data acquisition interval time to obtain a time interval matrix; Building a ground surface subsidence prediction model, and training the ground surface subsidence prediction model by adopting the time interval matrix and the time sequence InSAR subsidence data set to obtain a target model, wherein the target model is used for monitoring ground surface subsidence of a target area to be monitored. In one embodiment, the training the surface subsidence prediction model using the time interval matrix and the time series InSAR subsidence dataset comprises: Determining the time interval matrix and the time attention and the space at