BP神经网络在矿井突水水源识别中的应用

Application of BP Neural Network in Recognising Mine Water-bursting Source

针对矿井突水水源的水化学特征,采用Na^++K^+、Ca^(2+)、Mg^(2+)、Cl^-、SO_4^(2-)、HCO_3^-6种水化离子的浓度作为识别矿井水源依据;以35组水源样品作为训练样本,运用Matlab软件对网络进行训练,建立6×6×4的网络优化模型;使用构建的BP神经网络对4组待测样本进行识别,并与实际突水水源类别进行比对。应用结果表明:BP神经网络能够准确地识别矿井突水水源,可为防治矿井水害提供有力的保障。

For water chemical characteristics of mine water-bursting source, using Na^＋＋K^,Ca^2＋,Mg^2＋,Cl^-,SO4^2-,HCO3^- six hydrated ion concentration as the basis of mine water recognition; 35 groups of water samples were selected as the training sample, and used Matlab software to train the network, established a 6×6×4 network optimization model, used BP neural network to identify four groups of the sample to be tested, and compared with the actual water inrush source category. Application results show： BP neural network can accurately identify the mine water-bursting source and give full play to its advantages can provide a powerful guarantee for the prevention and control of mine water disaster.