矿区岩溶地表塌陷神经网络预测研究

Prediction of karst collapse in ming area based on neural network

针对近年来某矿岩溶地表塌陷频繁发生的现象,分析确定了影响地表塌陷的主要因素,构建了矿区岩溶地表塌陷预测BP神经网络模型,以训练后的BP网络模型对矿山帷幕注浆三期工程完成后可能形成的地表塌陷区的空间分布进行预测。并针对矿山现实塌陷情况,结合各区预测塌陷危险分级结果,提出了相应的岩溶地表塌陷灾害防治措施。实践表明,所建模型的预测结果与矿区地表塌陷实际情况相符,可为矿山后续帷幕注浆工程的设计与施工提供有益借鉴,为岩溶矿区地表塌陷灾害提供预警支持。

According to the frequent occurrence phenomenon of the karst surface subsidence in a mine in recent years,determined the main factors of surface subsidence,the BP neural network model of karst mining surface subsidence prediction was constructed.Using the trained BP network model,the spatial distribution of collapse area after the completion of phase III curtain grouting project was predicted.And according to the reality of mine collapse,combined with various predicted levels of collapse risk,the corresponding prevention measures were put forward for the surface karst collapse disaster.It is proved that the result of prediction was in line with the mining area actual situation.The model can provide a useful reference for the design and construction of post-grouting curtain surface subsidence and early warning for the karst mine surface subsidence.