Land subsidence (LS) refers to the problem of surface subsidence caused by man-made and natural causes. Natural factors mainly include natural compaction of loose strata and man-made factors mainly include human activities such as the exploitation of groundwater, oil and natural gas resources. Research and treatment of land subsidence have been a global problem(Dinar et al., 2021; Galloway et al., 2016; Zoccarato et al., 2018).In recent years, mining land subsidence (MLS) has received extensive attention and research. The mining of coal resources is of great strategic significance to China's long-term development. However, the excessive exploitation of underground coal resources has caused serious secondary disasters, which have had a huge impact on the landscape and ecological environment of the mining area(Zhang et al., 2019). Geological disasters such as surface deformation, fracture, and collapse caused by mining have destroyed a large number of farmland, vegetation, water resources and surface buildings, severely damaged the ecological environment of the mining area, and threatened the sustainable development of the coal industry(Akcin, 2021; Zheng et al., 2020). MLS destroys land resources, causing serious impacts on vegetation and water environment pollution (Ma et al., 2019; Zhao et al., 2020)and at the same time, it has also caused problems such as soil quality degradation and food security (Chen et al., 2019; Sun et al., 2019b). According to a survey conducted by the Bureau of Statistics of the People’s Republic of China, 92% of domestic coal production comes from underground mining. It is estimated that 10,000 tons of coal mining underground will cause about 0.2–0.33 hectares of ground surface subsidence. In China, the annual subsidence is expected to increase 2104 hectares (Hu and Xiao, 2013).Accurate prediction and evaluation of MLS is an important prerequisite for preventing mining subsidence damage, reducing subsidence disasters, and protecting the ecological environment, especially the areas with serious potential land subsidence hazards in China (Yuan et al., 2020).For a long time, methods such as probability integral method and influence function method, which focus on the analysis of mining subsidence law and the study of subsidence mechanism, have been the main methods of mining subsidence prediction, which provide theoretical support for subsidence control and disaster prevention.(Chen et al., 2016; Ju and Xu, 2015; Ma et al., 2017; Yan et al., 2018). With the rapid development of the subject of artificial intelligence, many scholars have cited artificial intelligence technology in the prediction and evaluation of MLS(Mohammady et al., 2021; Yan et al., 2021b).Research methods based on numerical simulation are also widely used in the evaluation and analysis of MLS (Liu et al., 2021; Sikora and Wesolowski, 2021). At present, the probability integral method is the most widely used subsidence prediction method in China (Xing et al., 2021; Yuan et al., 2020). The MLS is affected by mining conditions and geological conditions, and it is affected by a large amount of randomness and ambiguity(Li et al., 2021).It can be seen from the above-mentioned literature analysis that MLS is seriously harmful to the environment. There are many research methods for MLS, but they have the following shortcomings in the application:
(1) MLS is closely related to mining and geological conditions and is affected by strata movement and overlying rock activities. It is a complex problem of quantitative and qualitative multi-index coupling, with significant regional and particular characteristics. In the evaluation and analysis, it is difficult to comprehensively analyze complex indicators, and can only simplify the analysis of a large number of qualitative indicators. The traditional predictive evaluation methods established by simplifying complex influencing conditions are difficult to describe the impact of complex factors such as key strata, loose strata, and lithological coupling of different strata. Therefore, the final result often deviates from the actual situation. (2) In the past, few models can well solve the randomness and ambiguity problems in MLS research. Therefore, this study uses the cloud model to solve this long-term problem, based on the establishment of an index system.
Therefore, in view of the shortcomings of traditional evaluation and prediction models, such as difficulty in describing key strata, loose strata, lithological coupling of different strata, and some complex and fuzzy factors, this study adopts the safety evaluation method(Ranjan and Hughes, 2014; Sovacool et al., 2011; Vithayasrichareon et al., 2012) to integrate the quantitative dimensions (the occurrence conditions of the coal seam, mining design parameters), and the qualitative dimensions (the nature of the rock, the impact of key strata ) to construct the PSR (pressure-state-response) comprehensive conceptual model of MLS(Fang et al., 2008; Wang and Sun, 2017) based on the characteristics of coal mining, land subsidence, prediction and evaluation, etc. In addition, the comprehensive weighted cloud model of game theory weighted subjective weight and objective weight is introduced for analysis and evaluation, which improves the rationality and safety of mining subsidence evaluation.