BP神经网络在开采沉陷相似材料配比中的应用

Application of BP Neural Network in the Proportion of Similar Materials of Mining Subsidence

相似材料模型试验是矿山开采沉陷机理研究的重要研究方法,相似材料配比是实现相似材料模型模拟可靠性的关键因素。由传统试验方法确定相似材料配比费时、费力。综合分析了相似材料选择原则,以中国矿业大学研究制作的相似材料配比表为基础,建立了相似材料配比的BP神经网络模型。以33组试验数据作为训练和测试样本,模型预测的最大相对误差为7.39%。研究表明:所建立BP神经网络模型可基本反映出相似材料抗压、抗拉强度与各材料配比之间的内在影响规律,用该模型进行相似材料配比预测是可行的。

The similar material model test is one important research method to study the mining subsidence mechanism,and similar material proportion is the key factor to achieve a reliable similar material model.Similar material proportion confirmed by traditional test method is time-consuming and laborious.A BP neural network model for similar material proportion was established based on the table of similar material proportion.33 groups of experimental data were used as training and testing samples,and the maximum relative error of model prediction was 7.39%.The result shows that the BP neural network model can basically reflect the inner influence rules between similar materials＇ compression and tensile strength and their proportion.It is feasible to predict the similar materials proportion by this model.