摘要
矿井涌水量预测对矿山的安全生产和地下水资源的保护都有着重要意义。将广义回归神经网络(GRNN)引入到矿井涌水量预测中,以实例为研究基础,提出采用GRNN对矿井涌水量预测问题进行建模,将大气降水、采空区面积和底板构造断裂及采动裂隙三个影响因子作为网络输入,涌水量作为预测输出,采取交叉验证方法获得光滑因子来建立预测模型。预测结果表明,GRNN模型的预测值与真实值的最大相对误差仅为4.27%,而BP神经网络预测的最大相对误差为10.48%。同时,减少训练样本数量,即应用于小样本预测问题时,GRNN模型的预测结果较BP神经网络精度高且稳定性好。因此,应用GRNN模型进行矿井涌水量预测是准确的、可行的。
Prediction of water yield in mine plays an important role in mine safety production and protection of groundwater resource.The generalized regression neural network（GRNN） was introduced into forecasting of water yield in mine.Taking known mine data as practical example,the water yield in mine was predicted by GRNN.The meteoric water,goaf area,and plate structure fracture and mining-induced fracture were selected as input of the network model,obtaining smoothing factor through cross validation,the prediction model was constructed.The results of prediction showed that the maximum relative error between the predicted value with the real value was only4.27%,the maximum error of BP neural network was 10.48% in same condition.Meanwhile,under small samples conditions,GRNN prediction model also had good stability and high accuracy.So the GRNN prediction model is accurate and feasible for prediction of water yield in mine.
出处
《中国安全生产科学技术》
CAS
CSCD
2014年第11期90-93,共4页
Journal of Safety Science and Technology
基金
辽宁工程技术大学博士启动基金
关键词
矿井涌水量
预测
广义回归神经网络
water yield in mine
prediction
generalized regression neural network（GRNN）
