CAJ | 학술논문

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。
상사재료모형시험시광산개채침함궤리연구적중요연구방법,상사재료배비시실현상사재료모형모의가고성적관건인소。유전통시험방법학정상사재료배비비시、비력。종합분석료상사재료선택원칙,이중국광업대학연구제작적상사재료배비표위기출,건립료상사재료배비적BP신경망락모형。이33조시험수거작위훈련화측시양본,모형예측적최대상대오차위7.39%。연구표명：소건립BP신경망락모형가기본반영출상사재료항압、항랍강도여각재료배비지간적내재영향규률,용해모형진행상사재료배비예측시가행적。
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.