Genetic algorithm-based parameter inversion and pipeline subsidence prediction

Abstract

To invert mine subsidence parameters and predict the timing of associated risk to underground pipelines passing through mining areas with incoherent subsidence information, this study proposes a model based on combining the advantages of interferometric synthetic aperture radar (InSAR) technology, the probability integral method (PIM), segmented Knothe time function, and genetic algorithm. Firstly, InSAR technology is used to analyze a time series of surface deformation in a mining area, circle the range of large-gradient deformation, and distinguish the subsidence edge from the subsidence center. In the case of the subsidence edge (small-scale deformation), InSAR results are retained, whereas in the case of the subsidence center (large-scale deformation), subsidence parameters, such as the subsidence coefficient and influence radius, are obtained via the genetic algorithm. The subsidence of the working face in the mining area is determined by combining InSAR information and the PIM parameter inversion-based basin model. This allows us to obtain continuous surface subsidence information, study surface subsidence trends over time, and predict future pipeline subsidence. The researchers subsequently test the efficacy of this model in the Qinchi mining area in Shanxi Province, China. Sentinel-1A images were used as the data source, combine the PIM and segmentation time function, and use the genetic algorithm to obtain the subsidence mining parameters. The inversion results show that the subsidence coefficient is 0.5917 (with an error of 2%), and the main influence angle tangent value is 1.927 (with an error of 2.18%), indicating high inversion accuracy. Based on this model, the study predict that subsidence in the Qinchi mining area will threaten the structural integrity and operation of the intersecting pipeline by November 2025.