国内地下金属矿山采用无底分段崩落法采出的矿量占绝大多数。该采矿法工艺比较简 单,均采用截止品位放矿来控制出矿,此放矿方式存在着废石混入面大、混入机会多、矿石 贫化率高的缺陷,而贫化率高的结果造成矿山企业的一级产品质量下降,二级产品处理的费 用增加。因而此采矿方法的研究和应用在世界和国内的矿山都是一个关心的课题。随着科技 在矿业中的逐步应用,矿山一直在寻求用不同的方法来解决此问题,本报告以梅山铁矿作为 依托着重研究低贫化放矿工艺并用神经网络进行实时优化预测,确定合理的贫化率指标和其 他技术经济指标,来指导和加强放矿管理。所谓低贫化放矿即在保证矿山总体回收率基本不 变的情况下,逐步降低贫化率,提高出矿品位,从而降低废石混入,为企业带来效益。 神经网络是基于实例的研究方法,它对非线性系统有较好的逼近效果,它能总结归纳出 变量之间的关系,快速实时预测优化技术经济指标,它对指导生产,加强管理,具有重要意 义。因此,采用神经网络方法为优化预测低贫化放矿技术经济指标提供了一种新的研究方法, 它可以充分挖掘经验数据和信息,进行仿真优化组合,为低贫化放矿技术经济指标提供最优 的变化区域。 本文主要讲述了低贫化放矿的原理以及与截止品位放矿的区别,接着对神经网络的理论 进行了简介,通过对低贫化放矿的剖析,针对梅山建立了低贫化放矿的模型,通过神经网络 模型实时预NlJ出较优的贫化率,最后确定梅山铁矿实行低贫化放矿的具体定量指标。低贫化 放矿的推广应用对于实行无底分段崩落法的矿山具有一定的社会价值和经济价值。 关键词:贫化率截止品位低贫化神经网络Abstraet MostoforesinundergroundmetalminesinChinawereminedbythe 5111Pillar一freesubdriftingmethod.ThemethodhassimPleProeessand eontrolsoreremovalbyeut一offgradeo邝drawing.Ithasshorteomingsof mixtureoflargeamountofbarrenstones,manychaneesofmixtureof barrenstonesandhighoredilutionratio.Thedilutionratiomayresultin deereasingqualityoffirst一gradeProduetsofmineenterprisesand increasinghandlingeostsofsecond一gradeProduets.Therefore,thestudy andaPPlicationoftheminingmethodaretoPiesdrawingattentionof technieiansbothathomeandabroad.WithgradualapPlieationofscience andteehnologyintheminingindustry,theminesarelookingfordifterent methodstosolvetheProblem.BasedonMeishanIronMine,thedilution ratiodrawingProeesswasstudiedandreal一timeoPtimizationPredietion wasearriedoutbyneuralnetwork,50astodeterminereasonabledilutioh ratioindexandotherteehnicaleeonomieindexes,andtoguideand strengthenoredrawingmanagement.Thedilutionratiooredrawing15that underthecireumstaneeofunehargedoverallreeoveryofthemine,the dilutionratio15deereasedgraduallyandoreremovalgrade15inereased,50 astodeereasemiXtureofbarrenstonesandtobringbenefitforthemine. Theneuralnetwork15aresearehmethodbasedonexamPlesandhas betteraPProximatingeffectforthenon一linearsystem.Itcansummarize relationbetweenvariatesandtheteehniealeeonomicindexofquiek real一timePredietionoPtimization,andhasimPortantmeaningtoguide Produetio琢andstrengthenmanagement.Therefore,theneuralnetwork methodProvidesanewstu勿methodforPredictionoftechniealeeonomie indexofthedilutionratiooredrawing.ItmaysuffieientlydeveloP exPerieneedataandinformation,cari了outemulationandoPtimizationeombination,andProvideoPtimumchangingfieldfortheteehnieal eeonomieindexofthedilutionratiooredrawing. ThePrineiPleofthedilutionratiooredrawinganddiffereneebetween thedilutionratiooredrawingandtheeut一offoredrawingaremainly exPoundedandthetheoryoftheneuralnetwork15brieflyPresented. Throughanalysisofthedilutionratiooredrawing,themodelofthe dilutionratiooredrawingwasestablishedinMeishan.Inthe11ghtofthe neuralnetworkmodel,betterdilutionratiowasPredictedandaetual quantitativeindexofthedilutionratiooredrawingwasdeterminedin MeishanIronMine.PoPulationandapPlicationofthedilutionratioore drawingmethodhaseertainsocialvalueandeeonomicvaluefor imPlementationofthe5111Pillar一freesubdriftingmethod. KeywordsDilutionratio,Cut一offgrade,Lowdilution,Neuralnetwork目录 l课题研究的来源及选题的主导思想...........................................................l 1.1研究课题的提出.…,...............................1 1 . 1.1实行低贫化放矿的重要性........................,二,......,............1 1.1.2实行低贫化放矿的必要性.........................,............................1 1.1.3实行低贫化放矿的安全性................................................,.......2 1.2报告选题的主导思想.…,............................,2 1.3国内外发展概况和本研究的主要工作..................................................4 1 . 3.1国内外发展概况....................................................4 1.3.2本报告研究的主要工作..............................44 2低贫化放矿规律研究二,.....,....................46 2.1现行截止品位放矿理论研究..............................................................6 2,l‘2 截」_卜品位放矿原理 截止品位存在问题 2.2覆盖岩层‘卜矿岩移动规律的研究........................................................10 2 . 2 . 1无底柱分段崩落法放矿贫化特点.....................,................,10 2‘2.2覆盖岩层下放矿散体颗粒的移动特点...............................................12 2.2.3放矿过程分析............................................................................13 2.2.4矿岩移动控制对放矿品位的影响...................,...........................14 2.3低贫化放矿理论,.................................................,....................巧 2 . 3.1低贫化放矿原理....................................................,...............巧 2.3.2低贫化放矿与截止品位放矿的区别二,.......................,............,.…18 2.3.3低贫化放矿的优势…,...................................................................19 3梅山铁矿低贫化放矿模拟试验研究.......,..................,.二,..............20 3.1梅山铁矿概况.......................、..............................................,.…20 3.2低贫化放矿_〔艺在梅山铁矿应用可能性分析.........................,.........21 3.3物理模型模拟试验...............................................,..............22 3 . 3 . 1实验室出矿控制...........................................................,........22 3.3、2试验方法及试验结果....................,.......................,.,.........23 3.3.3试验结果分析...........,..................................,.................31 3 . 4小结.......,...........................................................,...........334神经网络在低贫化放矿工艺中实时优化预测技术经济指标.............................34 4,1影响低贫化的因素分析........................,.....................................34 4.2神经网络拓扑结构................................................................,...35 4 . 2.1前向网络(前馈网络).…,............................................................35 4 . 2.2层内有反馈的前馈网络...............................,.................二35 4 . 3前向网络的模糊学习机制,................................,.........................36 4.3.1学习准则................................................................................37 4.3.2模糊学习方法的实现步骤.,.,几二‘,.......,.…,...........................38 4.4神经网络预测优化模型构造,.,............................................,..38 4.5神经网络计算机实现.........,.............................................44...,4…39 4.6对主要因素进行仿真和优化分析....................,............................41 4.6.1预测一、二分层回收率技术经济指标...,.....................................41 4 . 6 . 2贫化率对〔技术经济指标3]的影响..…,................,.......................43 4 . 7结论......,...............................................................................45 5放矿管理与贫化控制措施..........................................,......................46 5 . 1健全合理的放矿管理制度............................,..‘.....................46 5.2出矿品位监测与放矿截止品位控制措施............................,…,.......47 6结束语...................,...........................................................,...49 6.1梅山铁矿实行低贫化放矿的可行性.........,........,......................49 6.2低贫化放矿在国内的应用范围以及应用中注意的问题.............................49 6.2.1低贫化放矿工艺的适用条件..............................................、…49 6 . 2.2矿山在开展低贫化放矿试验及工业应用时应注意以下儿个问题..............50 6 . 3低贫化的推广价值..,............................一50 6 . 3.1.低贫化放矿卜艺的优势...............................、、..…,、.…,、、、…、..…50 6 , 3.2.低贫化的推广价值.........................51 6 . 4神经网络实时优化预测的优越性........................................................51 致谢............,.....,.......................................................................53 附录l:攻读硕十学位期间发表的文章.........................................................54 附录2: