The invasive weed optimization–based inversion of parameters in probability integral model

Jingyu Yang,Chao Liu,Tianyang Chen,Yaming Zhang

DOI: https://doi.org/10.1007/s12517-019-4592-9

2019-07-01

Arabian Journal of Geosciences

Abstract:The probability integral model is popularly used to study and predict the movement and deformation of ground surface caused by the coal mining. The parameters to be inversed in this model mainly impact the accuracy in prediction. To inverse the parameters, the intelligent optimization methods (IPO) are a set of effective methods adopted in coal mining. In this paper, we originally innovate the invasive weed optimization (IWO) to the parameter inversion in this model. In contrast to the particle swarm optimization and genetic algorithm, it shows a more robust, higher accuracy, and self-adaptive performances in the simulated situations and in an actual case in Inner Mongolia. Moreover, we also found that the intensive fluctuation of the ground surface would negatively impact on the accuracy of the parameters inversed. In the case in Inner Mongolia, even though this negative effect is stronger than that caused by the selection of optimization methods, IWO could partially compensate this impact more than others. In terms of this point and its performance in the comparative experiments of simulated data, IWO is assumed as a better solution to inverse the parameters in the probability integral model comparing with the other commonly used methods in coal mining.

geosciences, multidisciplinary