基于BP神经网络的露天边坡角预测.docVIP

PAGE 6 PAGE 1 基于BP神经网络的露天边坡角预测 张儒1，陈新2 （1.广东核力工程勘察院，广东 广州 5108002；2.中南大学 资源与安全工程学院，湖南 长沙 410083） 摘 要：针对广东泥竹塘铁矿露天边坡稳定性问题，为了获得矿山的稳定露天边坡角，研究使用BP神经网络进行预测分析。在预测过程中，以边坡岩体质量系数、岩体综合抗压、抗拉强度、内聚力、结构面力学特性、边坡高度及岩体密度等7个指标为输入因子，综合国内矿山27组露天矿山现场数据，建立网络学习、训练样本库，从而实现泥竹塘铁矿稳定露天边坡角的预测。结果表明：露天边坡角的BP神经网络预测模型最大误差小于3%，训练输出误差较小，精度较高，得到的泥竹塘露天铁矿上盘最终边坡角的预测值为42.8°，上盘最终边坡角的预测值为40.1°。多年的生产实践表明，该预测成果与实际基本相符，可为今后类似工程提供参考。 关键词：边坡角；稳定性；BP神经网络；预测 Prediction of Slope Angle based on BP Neural Network ZHANG Ru1, CHEN Xin2 (1.Guangdong Nuclear Force Institute Of Engineering Investigation, Guangzhou Guangdong 510800, China; 2. School of Resources and Safety Engineering, Central South University, Changsha, Hunan 410083, China) Abstract： Aiming at solving the stability problem?of slope in?Guangdong NiZhutang?iron ore,?this paper use BP?neural network prediction?analysis to?obtain?the?stability slope angle.?In the prediction process, the?slope rock mass?quality coefficient,?comprehensive compressive strength,?tensile strength,?cohesion,?surface characteristics,?slope height and?mass?density?was chosen as input factors,?and comprehensively consider the field data of?27 groups domestic mine open-pit mine. A BP neural network was establish to predict the stability?slope angle in NiZhutang?iron ore.?The results show that: the?maximum error of prediction model is less than 3%, which means that the training output?error of BP neural network?is small. The value slope angle?of hanging?side in?NiZhutang?Open-pit Mine is forecasted 42.8?degrees,?the value angle?of ultimate slope?is 40.1 degrees.?Many years of production practice shows that the?forecast?results?are basically consistent with the actual situation, and this results can provide?a reference for similar projects. Key words: slope angle; stability; BP neural network; prediction 引言 露天边坡角的确定是边坡工程的稳定性研究的关键问题，影响到矿山的经济、安全生产[1-2]。边坡角的大小选取是一把双刃剑，取值偏大可以减少生产成本，但随之而来的增大了发生地质危害的可能性，反之取值偏小虽然减少了地质灾害发生的几率，却增大了生产成本[3-4]。广东泥竹塘铁矿随着开挖深度的增加，边坡高度逐渐增加，边坡稳定性及其安全隐患逐步出现。因此，采用科学合理