Subsidence prediction serves as a crucial risk analysis and management tool in mining
regions. To mitigate risks associated with ground subsidence, understanding not only
the magnitude but also the relative position of mining-induced ground movements and
deformations is essential. The source of subsidence stems from the convergence of an
underground void due to the overlaying rock mass and related pressure. This is
transferred to the surface, resulting in a typical subsidence trough. In empirical
functional prediction methods, this shape is modeled by an influence function.
Classical subsidence prediction offers an efficient solution for symmetrical shapes and
Gaussian-distributed subsidence; however, real-world observations reveal
asymmetrical and uniquely shaped patterns. Various mathematical approaches have
been implemented to account for these patterns to improve subsidence prediction.
Nonetheless, they possess significant disadvantages such as complexity, noninterpretability of parameters, and an inability to accommodate other patterns. This
article introduces a novel solution for subsidence prediction, addressing both
asymmetry and shape deviations concurrently or independently, while retaining
compatibility with classical solutions.
To evaluate the prediction method, the best-estimated parameters are applied across
different scenarios, including a full case study of subsidence above energy storage salt
caverns in the Middle European region. The application of the new solution significantly
improves the subsidence prediction accuracy, with up to a 25% reduction in mean
square error compared to the classical subsidence prediction method and up to a 12%
improvement over individual pattern approaches