Prediction model with multi-point relationship fusion via graph convolutional network: A case study on mining-induced surface subsidence

PLOS ONE RESEARCH ARTICLE Prediction model with multi-point relationship fusion via graph convolutional network: A case study on mining-induced surface subsidence Baoxing Jiang ID1,2,3☯, Kun Zhang ID1,2☯*, Xiaopeng Liu1,2, Yuxi Lu2,4 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 1 State Key Laboratory of Mining Response and Disaster Prevention and Control in Deep Coal Mines, Huainan, China, 2 School of Geomatics, Anhui University of Science and Technology, Huainan, China, 3 School of Information Science and Engineering, Shandong Normal University, Jinan, China, 4 Faculty of Land Resources Engineering, Kunming University of Science and Technology, Kunming, China ☯ These authors contributed equally to this work. * Abstract OPEN ACCESS Citation: Jiang B, Zhang K, Liu X, Lu Y (2023) Prediction model with multi-point relationship fusion via graph convolutional network: A case study on mining-induced surface subsidence. PLoS ONE 18(8): e0289846. https://doi.org/ 10.1371/journal.pone.0289846 Editor: Praveen Kumar Donta, TU Wien: Technische Universitat Wien, AUSTRIA Received: April 14, 2023 Accepted: July 27, 2023 Published: August 16, 2023 Copyright: © 2023 Jiang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting information files. Funding: 1. State Key Laboratory of Mining Response and Disaster Prevention and Control in Deep Coal Mines, SKLMRDPC21KF19 Mr. Kun Zhang 2. Department of Science and Technology of Anhui Province, 2022h11020024 Mr. Kun Zhang 3. Department of Science and Technology of Anhui Province, 2108085QE207 Mr. Xiaopeng Liu 4. Science and Technology Bureau of Huainan City, Accurate prediction of surface subsidence is of significance for analyzing the pattern of mining-induced surface subsidence, and for mining under buildings, railways, and water bodies. To address the problem that the existing prediction models ignore the correlation between subsidence points, resulting in large prediction errors, a Multi-point Relationship Fusion prediction model based on Graph Convolutional Networks (MRF-GCN) for mining-induced subsidence was proposed. Taking the surface subsidence in 82/83 mining area of Yuandian No. 2 Mine in Anhui Province in eastern China as an example, the surface deformation data obtained from 250 InSAR images captured by Sentinel-1A satellite from 2018 to 2022, combined with GNSS observation data, were used for modeling. The deformation pattern of each single observation point was obtained by feeding their deformation observation data into the LSTM encoder, after that, the relationship graph was created based on the correlation between points in the observation network and MRF-GCN was established. Then the prediction results came out through a nonlinear activation function of neural network. The research shows that the R2R2 value of MRF-GCN model was 0.865 0, much larger than that of Long-Short Term Memory (LSTM) and other conventional models, while mean square error (MSE) of MRF-GCN model was 1.59 899, much smaller than that of LSTM and other conventional models. Therefore, the MRF-GCN model has better prediction accuracy than other models and can be applied to predicting surface subsidence in large areas. 1 Introduction Surface subsidence is a common geological environmental disaster, which is increasingly concerned [1–4]. It is characterized by a long development cycle, irreversibility, and continuous accumulation of destructive effects, that usually brings a series of unfavorable impact on people’s production and daily life [5–9]. Especially surface subsidence caused by underground PLOS ONE | https://doi.org/10.1371/journal.pone.0289846 August 16, 2023 1 / 17 PLOS ONE 2021130 Mr. Kun Zhang 5. Anhui University of Science and Technology, QNYB2021-02 Mr. Kun Zhang 6. Anhui University of Science and Technology, 2023yjrc43 Mr. Kun Zhang 7. Huaibei Mining (Group) Co., 2022101 Mr. Kun Zhang 8. Huaibei Mining (Group) Co., 2023067 Mr. Kun Zhang 9. Anhui Construction Engineering Group, SG2025Q11 Mr. Kun Zhang The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. Prediction model with multi-point relationship fusion via graph convolutional network coal mining may lead to irreparable damage to buildings, roads and farmland, etc. in mining areas [10, 11]. Prediction with high accuracy contributes to the prevention and control of the damage caused by disaster such as surface subsidence in large-scale regions. The existing models for predicting surface subsidence are mainly divided into two types: physical models and statistical models [12–14]. Physical models [15–18] are established according to the mechanical properties of different rock layers to predict the surface subsidence, of which the limitation is that the strong dependence on the characteristics of geological structure in the study area what makes it difficult for them to form a unified and universal framework. Statistical models [14, 19] obtain regular patterns through the analysis of a large amount of existing data, and based on this, predict the future trend of the data. The statistical model requires the acquisition of strict sedimentation data, which has very high requirements on the acquisition of data, especially in the past, when manual observation was time-consuming, laborious, and inefficient, making it difficult to obtain data and predict surface subsidence in large areas. With the development of remote sensing technology [20–22] and deep learning [23], acquisition and processing of long-term, large-scale surface deformation data was no longer a problem. But there is an increasing demand for data mining accuracy. The emergence of Graph Convolutional Network (GCN) [24–26] is well suited to meet the requirements of accurate data mining. GCN is an practical variant of the traditional convolutional neural network (CNN) [27] proposed by Kipf and Welling [24]. GCN can effectively perform convolution operations on irregular topological graph structure information, thereby learning the hidden information associated with the graph interior. Taking surface subsidence as an example, each observation point does not exist independently, and there are strong internal connections between them. Based on these connections, the pattern of surface subsidence can be obtained. In general, the graph structure data consists of a set N of nodes and a set E of edges, where N 2 Rn�m stores the n nodes of the graph structure, and the information of each node is uniquely represented by an m-dimensional vector, and the set of edges E 2 Rn�n stores the connectio (...truncated)