基于灰色神经网络组合模型的矿区地表变形预测方法研究

Prediction Method of Surface Deformation in Mining Area based on Grey Neural Network Model

矿山开采区地表变形是资源开采引起的重要现象。监测、分析其内在规律,建立预测模型预计开采区地表沉降,对矿山安全生产和企业经济可持续发展具有重要意义。针对地表沉降变形预测,分别建立了BP人工神经网络模型、灰度预测模型、时间序列模型以及灰色神经网络组合模型,探讨四种模型适用场景及模型局限性,结合矿区地表一年的实际监测值进行模型精度评定,分析比较四种模型的模型特点以及预测结果。综合比较表明组合模型融合了BP神经网络与灰度预测模型的优势,预测精度高,更加贴合实际测量值,且广泛适应性强,可适用于多种应用场景,能较好地反映矿区地表沉陷变形发展趋势。

The surface deformation in mining area is an important phenomenon caused by resource exploitation.It is of significance for mine safety production and sustainable development of enterprise economy to monitor and analyze its internal law and build prediction model to predict the surface subsidence of mining area.Aiming at the prediction of surface subsidence and deformation,this paper builds BP artificial neural network model,gray prediction model,time series model and gray neural network combination model respectively,discusses the applicable scenarios and limitations of the four models,evaluates the accuracy of the model in combination with the actual monitoring value of the mine surface in one year,analyzes and compares the characteristics and prediction results of the four models.A comprehensive comparison shows that the combined model combines the advantages of BP neural network and gray prediction model,has high prediction accuracy,is more in line with the actual measurement values.This model has wide adaptability,can be applied to a variety of application scenarios,and can reflect the development trend of surface subsidence deformation in mining area.