截止品位与入选品位智能优化的理论与方法研究
|
摘要
矿山采选系统的各个阶段需要对矿石品位进行控制。矿石品位包括地质品位、边界品位、截止品位、原矿品位、入选品位及精矿品位等，涉及到地质勘探、采矿、选矿等多个部门，由于这些品位的选取涉及到许多因素，诸如矿床地质条件、开采技术条件以及选冶技术经济条件等等，因此众多专家学者不断探索科学合理的优化决策方法，其主要目标是：资源充分利用，矿山收益最大，而实现这两个优化目标的关键因素是：放矿截止品位和入选品位。放矿截止品位是无底柱分段崩落法放矿过程中最后一次(当次)放出矿石的品位。截止品位低了，矿石贫化大，会增加矿石处理成本，同时在选矿加工生产能力一定的情况下会使精矿量减少；截止品位高了，损失率提高，不仅使可利用的矿产资源永久损失，增大基建成本，而且还使企业对矿石的处理能力得不到充分利用，矿山寿命缩短，同样会影响企业的经济效益。因此，合理确定的截止品位关系到企业眼前的经济效益和资源的可持续性利用。矿石的入选品位在采选过程中具有承上启下的作用，既取决于矿体平均品位以及采矿过程中的截止品位和贫化率，又影响选矿回收率和精矿品位等选矿指标，最佳的入选品位有利于提高矿山的整体效益和矿产资源的利用率。
由于生产环境的特殊性、放矿方式的高危性和矿山经营的粗放性，对于采用无底柱分段崩落法放矿的矿山企业，其放矿截止品位和入选品位一般由实验数据和现场放矿人员的经验来确定。随着资源的日趋紧缺以及科技水平、管理水平的提高，人们逐渐认识到，这种基于经验的截止品位和入选品位虽然简单易行，但大大地增加了采选成本，严重地造成了资源的浪费。随着矿山地质品位差异以及生产技术、管理水平、市场经济的变化，非常有必要建立截止品位、入选品位与经济效益、资源利用效益等相关的数学模型与系统，对截止品位和入选品位进行跟踪优化。
选题来源于武汉钢铁集团矿业有限责任公司大冶铁矿委托项目“多金属矿截止品位优化及其生产动态管理研究”，主要研究两个问题：1．如何确定放矿截止品位和入选品位，既能满足确定的精矿品位，又能使整个采、选过程的收益最大，且兼顾到资源的持续利用?2．当最优截止品位和最佳入选品位给定后，如何动态控制采出矿量、入选矿量和精矿量及相应品位?主要研究多年来困扰金属矿山截止品位优化难、入选品位动态管理难的问题，为矿山企业提供解决这一系列问题的科学方法和手段，从而达到提升矿产资源利用率，提高企业经济效益的目的。
本文主要研究人工神经网络、模糊系统、进化计算等智能方法对采选过程的两个关键品位指标(截止品位与入选品位)进行优化的理论与方法，并以大冶铁矿进行实证研究。其主要内容包括：
(1)提出对一时间段内的截止品位进行数值模拟的方法，采用粒子群算法对损失率函数模型进行结构辨识和参数估计，构建截止品位与损失率之间的函数关系式。在生产报表中没有截止品位α_j的历史记录，不可能直接通过学习获取表达式φ(α_j)。从理论上看，每个月的截止品位α_j在一定范围内服从正态分布；损失率φ和截止品位α_j呈正向关系，即截止品位越大，损失率越大，反之亦然。可认为，截止品位α_j服从中心为u=18，宽度σ=1的正态分布密度函数为f(a_j)=e~((a_j-18)~2／2)，提出截止品位的数值模拟算法，给出了其算法的收敛性证明，并成功模拟得到了大冶铁矿2005年1月至2007年11的截止品位值。将粒子群优化算法用于损失率与截止品位数学模型的结构辨识和参数估计中，计算得到损失率函数为φ=1.6508α_j-0．1175，为截止品位和入选品位优化工作奠定了基础。
(2)将进化计算与神经网络进行嵌套，提出了截止品位与入选品位的进化—神经集成的优化模型。截止品位及入选品位的优化是建立在净现值最大化的基础上的，考虑利润，成本，储量和各种品位之间的关系，建立非线性模型，其智能优化的主要内容包括：①数值模拟，其功能是获取截止品位α_j与损失率φ的数据关系{(α_j~i，φ~i)}；②BP神经网络，主要计算截止品位α_j、入选品位α_r、地质品位α_t、可采储量q_t与选矿金属回收率ε的关系ε=ε(α_t，q_t，α_j，α_r)；③模糊系统，用于获取截止品位α_j、入选品位α_r、地质品位α_t、可采储量q_t与采选总成本C的模糊规则，即用一组模糊规则描述关系式C=C(α_t，q_t，α_j，α_r)；④进化—神经集成。研究结果表明：现在大冶铁矿截止品位18％，入选品位41-42％的生产方案有待改进，当截止品位为17.8337-17.8367％，入选品位取值为46.4％，2007年1月至2007年11月的精矿量增加139200吨，净现值增加6.698百万元。将粒子群-神经优化算法(PSO-ANN)与遗传-神经优化算法(GA-ANN)及模拟退火-神经优化算法(SA-ANN)进行了比较，研究证明粒子群-神经优化算法性能最佳，并设计了简易的品位优化输入—输出界面。
(3)考虑铁、铜两种金属，采用成本分块的思想，构建了利润最大化的非线性规划模型，将粒子群算法与神经网络嵌套构成粒子群-神经模型(PSO-ANN)来优化截止品位α_j和入选品位α_r。具体操作为：用截止品位与入选品位一起组成进化计算的粒子群个体，用自适应神经网络建立收益(适应度函数)与各粒子的局部联系，然后利用粒子群算法的全局搜索功能找出使适应度函数(收益)最大时的品位组合。2008年考核计划的可采地质储量为153.17万吨，其地质品位为TFe52.60％，Cu0.306％，其最优截止品位和入选品位分别为14.6848％，入选品位为42.1388％。
(4)兼顾经济效益和资源利用效益，建立截止品位和入选品位的多目标优化模型。首先采用神经网络建立以截止品位和入选品位为变量，精矿量、净现值和资源利用率为目标的函数，再对其方案进行模糊综合评价，将得到的模糊隶属度加权值作为遗传算法的适应度函数，全局搜索出使适应度函数最大，即最优的品位指标组合，实现截止品位和入选品位的动态优化，可为矿山企业提供更为科学合理的决策。
(5)将整个采选系统划分为三个阶段，并建立关键指标(矿量、品位)之间的函数关系式，根据确定的截止品位和入选品位，分别计算各个阶段的矿量和品位，对生产过程的各个环节进行控制和管理。采用截止品位为17．83％，入选品位为46．4％来指导2008年的采选矿生产时，①其采矿阶段，全年全矿损失率为18．31％，贫化率为22．17％，采山矿量为125．123万吨，混岩量35．6415万吨，毛矿量为160．7645万吨，毛矿品位为40．94％，可以通过毛矿量等指标来控制放矿，指导采矿生产；②在配矿阶段，要求入选品位在46．4％左右，假定抛废过程中自产矿混岩量全部抛出，外购矿品位为35％，则需外购矿量80．8278万吨；③选矿阶段全年入选矿量为241．5923万吨，选比在1．9877左右，全年总采出精矿量121．5436万吨，精矿品位在64．5％左右。
Every period of mining and milling in mine system should control the grades of ore. Thegrades of ore contain geological grade, the boundary grade, cut-off grade, grade of crude ore,concentrate grade and so on, and they relate to geological exploration, mining, milling and manyother departments, involving many factors, such as the geological conditions, mining technicalconditions and technical and economic conditions of milling and smelting. Many experts andscholars continue to explore the scientific method for decision-making, the main objectives are tomake full use of resources and to maximize the revenue, which need to find out the optimalcut-off grade and grade of crude ore. Cut-off grade is the grade of ore in the last time of oredrawing during sublevel caving with no sill pillar. If the cut-off grade is low, the dilution rate ofore will be high. This will increase the cost of ore processing, and will decrease the amount of theconcentrate in the condition of a certain mineral processing ability. Contrarily, high cut-off gradewill not only lead to waste of mine resource, but also increase the cost of fundamentalconstruction. Grade of crude ore plays a role as a link between mining and milling. It isdetermined by the average grade of ore bodies, cut-off grade and dilution rate. And it affects themilling recovery and concentrate grade. The optimum grade of crude ore can improve the benefitof mine and make use of mine resource. Therefore, reasonable determination of cut-off grade andgrade of crude ore is crucial to the economic benefit of enterprise and sustainable utilization ofresource, and it has important theoretical and practical meaning.
Because of the particularity of production environment, multiplicity of ore drawing andextension of mine management, in mine enterprise with sublevel caving with no sill pillar,generally, cut-off grade and grade of crude ore are determined by experiment data or worker'sexperience. Along with the increase of resources' scarcity, enhancement of technical andmanagement level, the experts and managers gradually realize that, the determining methodbased on workers' experience, easy and feasible though, greatly increases the mining and millingcost, and wastes resource seriously. With the difference in geological grade and the improvementin production technology, management technology and market economic condition, it is verynecessary to set up mathematical models and relative systems from cut-off grade and grade ofcrude ore to economic benefit and utilization benefit of resource, in order to dynamicallyoptimize cut-off grade and grade of crude ore.
This dissertation is supported by the Project "Research on the optimization of cut-off gradeand production dynamic management in metal mine", which originated from Daye Iron Mine inWuhan Iron and Steel Group Co. Ltd. It mainly resolves two questions. The first one is that, howto determine the cut-off grade for drawing and grade of crude ore, which meet the givenconcentrate grade, not only make the revenue of mining and milling maximum, but also take intoaccount the sustainable utilization of resource? The second one is that, when the optimal cut-offgrade and grade of crude ore are given, how to dynamically control the amounts and grades ofmining ore, milling ore and concentrate ore? So this research aims to resolve the problem ofoptimization and management of cut-off grade and crude ore in metal mine, to provide thescientific method and means for solving the series of questions, in order to upgrade the utilizationof mineral resources, and improve enterprises' economic benefit.
This dissertation mainly researches on the theories and methods of intelligent optimizing twokey grades, namely cut-off grade and grade of crude ore in mining and milling system, byartificial neural network(ANN), fuzzy systems(FS), evolutionary computation(EA) and so on,and take Daye Iron Mine as case studies. Its main contents are as follows.
(1) Propose a novel method that uses numerical simulation to acquire the value of the cut-offgrade in a certain period, use particle swann optimization for structure identification andparameter estimation of the function of loss rate, then construct the function relationship ofcut-off grade and loss rate. Due to the lack of record data of cut-off gradeα_j in practicalproduction, we can't acquire expressionφ(α_j).We know from the apriori information: Thecut-off grade value is normally distributed in certain universe of discourse; the greater cut-offgrade is, the greater loss rate will be, and vice versa. It can be concluded that the distributionfunction of cut-off grade value is f(α_j)= 1/(2π)e~((a_j-18)~2)/2 , where, the mean is u = 18, variance isσ= 1. The numerical simulation algorithm of cut-off grade is proposed, the convergence proofis given, and we successfully obtain cut-off grade values from January 2005 to November 2007in Daye Iron Mine. The Particle Swarm Optimization is used for structure identification andparameter estimation of math model between loss rate and cut-off grade, and the loss ratefunction isφ=1.6508α_j -0.1175, which is the basis of optimizing cut-off grade and grade ofcrude ore.
(2) Evolutionary algorithm and neural network are nested to be a EA-ANN integrated model,and use it to optimize the cut-off grade and crude ore grade. Optimal cut-off grade and grade ofcrude ore are based on maximizing the net present value (NPV). Considering the relationshipamong profit, cost, geological reserves and all kinds of grades, we establish a nonlinear model.The main four parts of intelligent optimization are as follows. The first one is data simulation,which is used to determine relationship between cut-off gradeα_j and loss rateφ; the second isBP neural network, which is to determine relationship between the milling recovery rateεandcut-off grade a_j, crude ore gradeα_r, geological gradeα_t , geological reserves q_t ; the third one is fuzzy system, which is used for determining relationship between the cost of mining & milling Cand cut-off gradeα_j, crude ore gradeα_r, geological gradeα_t, geological reserves q_t,; the lastone is the integration of EA and ANN. The result shows that, during the period of January toNovember in the year 2007, the optimal cut-off grade is 17.8337-17.8367%, and optimal grade ofcrude ore is 46.4%. Comparing with the present scheme (cut-off grade is 18%, grade of crude oreis 41-42%), the optimized scheme can increase the amount of concentrate by 139200 tons, andimprove the net present value by 6.698 million Yuan. Compared PSO-ANN with the GA-ANNand SA-ANN algorithms, it demonstrates that the performance of PSO-ANN is obviouslysuperior to GA-ANN and SA-ANN. A simple input-output interface of grades optimization isdesigned.
(3) Take account into the iron and copper, put total cost into several parts, construct thenonlinear optimization model to maximize the profit. An integration of PSO and ANNs to be aPSO-ANN integrated model, to optimize the cut-off gradeα_j and crude ore gradeα_r. Thedetails are as follows: Join cut-off grade and grade of crude ore together as particle of swarm forevolutionary computation; Make self-adaptive neural network to get the local connectionbetween the income value (fitness function) and each particle; Use PSO algorithm globally tosearch the optimal cut-off grade and grade of crude ore to maximize fitness function. Theproduction plan of Daye Mine in the year 2008 is that, geological reserves is 1.5317 million ton,geological grades are 52.60% and 0.306% for Fe and Cu respectively, through calculation, theoptimal cut-off grade and optimal grade of crude ore are 14.6848%, and 42.1388%, respectively.
(4) Balancing between economic benefit and resource utilization benefit, set up multi-objectiveoptimization model of cut-off grade and grade of crude ore. Neural networks are used toconstruct mapping relationship from cut-off grade, grade of crude ore to the amount ofconcentrate, total present value and total utilization rate; then make fuzzy comprehensiveevaluation of grade combinations, put the weighted fuzzy membership value as fitness functionof genetic algorithm, finally, globally search the best grades combination that make the fitnessfunction maximum, to research dynamic optimization of cut-off grade and grade of crude ore, toassist decision-making and management for mine enterprise.
(5) Divide mining and milling system into three stages, and establish functions of key indexes(amount and grade of ore). Given a certain combination of cut-off grade and grade of crude ore,we can calculate the amount of ore and the grade of ore of every stage, then we can control andmanagement each process of production. If we adopt cut-off grade 17.83%, grade of crude ore46.4% to lead practical production in the year 2008, it can be concluded as follows. In the miningstage, the average loss rate of the whole year is 18.31%, the dilute rate is 22.17%, the amount ofore recovery is 1.25123 million ton, the amount of mixed rock is 356.415 thousand ton, the totalamount and average grade of ore mining are 1.607645 million ton and 40.94%, respectively. Inthe mixing stage, if the grade of ore that should buy from outside is 35%, the amount of ore that need buy is 808.278 thousand ton. In the milling stage, the amount of ore milling is 2.415923million ton, the proportion of the amount of milling ore and concentrate ore is around 1.9877, theamount of concentrate of the whole year is 1.215436 million ton, and the grade of concentrate isaround 64.5%.
由于生产环境的特殊性、放矿方式的高危性和矿山经营的粗放性，对于采用无底柱分段崩落法放矿的矿山企业，其放矿截止品位和入选品位一般由实验数据和现场放矿人员的经验来确定。随着资源的日趋紧缺以及科技水平、管理水平的提高，人们逐渐认识到，这种基于经验的截止品位和入选品位虽然简单易行，但大大地增加了采选成本，严重地造成了资源的浪费。随着矿山地质品位差异以及生产技术、管理水平、市场经济的变化，非常有必要建立截止品位、入选品位与经济效益、资源利用效益等相关的数学模型与系统，对截止品位和入选品位进行跟踪优化。
选题来源于武汉钢铁集团矿业有限责任公司大冶铁矿委托项目“多金属矿截止品位优化及其生产动态管理研究”，主要研究两个问题：1．如何确定放矿截止品位和入选品位，既能满足确定的精矿品位，又能使整个采、选过程的收益最大，且兼顾到资源的持续利用?2．当最优截止品位和最佳入选品位给定后，如何动态控制采出矿量、入选矿量和精矿量及相应品位?主要研究多年来困扰金属矿山截止品位优化难、入选品位动态管理难的问题，为矿山企业提供解决这一系列问题的科学方法和手段，从而达到提升矿产资源利用率，提高企业经济效益的目的。
本文主要研究人工神经网络、模糊系统、进化计算等智能方法对采选过程的两个关键品位指标(截止品位与入选品位)进行优化的理论与方法，并以大冶铁矿进行实证研究。其主要内容包括：
(1)提出对一时间段内的截止品位进行数值模拟的方法，采用粒子群算法对损失率函数模型进行结构辨识和参数估计，构建截止品位与损失率之间的函数关系式。在生产报表中没有截止品位α_j的历史记录，不可能直接通过学习获取表达式φ(α_j)。从理论上看，每个月的截止品位α_j在一定范围内服从正态分布；损失率φ和截止品位α_j呈正向关系，即截止品位越大，损失率越大，反之亦然。可认为，截止品位α_j服从中心为u=18，宽度σ=1的正态分布密度函数为f(a_j)=e~((a_j-18)~2／2)，提出截止品位的数值模拟算法，给出了其算法的收敛性证明，并成功模拟得到了大冶铁矿2005年1月至2007年11的截止品位值。将粒子群优化算法用于损失率与截止品位数学模型的结构辨识和参数估计中，计算得到损失率函数为φ=1.6508α_j-0．1175，为截止品位和入选品位优化工作奠定了基础。
(2)将进化计算与神经网络进行嵌套，提出了截止品位与入选品位的进化—神经集成的优化模型。截止品位及入选品位的优化是建立在净现值最大化的基础上的，考虑利润，成本，储量和各种品位之间的关系，建立非线性模型，其智能优化的主要内容包括：①数值模拟，其功能是获取截止品位α_j与损失率φ的数据关系{(α_j~i，φ~i)}；②BP神经网络，主要计算截止品位α_j、入选品位α_r、地质品位α_t、可采储量q_t与选矿金属回收率ε的关系ε=ε(α_t，q_t，α_j，α_r)；③模糊系统，用于获取截止品位α_j、入选品位α_r、地质品位α_t、可采储量q_t与采选总成本C的模糊规则，即用一组模糊规则描述关系式C=C(α_t，q_t，α_j，α_r)；④进化—神经集成。研究结果表明：现在大冶铁矿截止品位18％，入选品位41-42％的生产方案有待改进，当截止品位为17.8337-17.8367％，入选品位取值为46.4％，2007年1月至2007年11月的精矿量增加139200吨，净现值增加6.698百万元。将粒子群-神经优化算法(PSO-ANN)与遗传-神经优化算法(GA-ANN)及模拟退火-神经优化算法(SA-ANN)进行了比较，研究证明粒子群-神经优化算法性能最佳，并设计了简易的品位优化输入—输出界面。
(3)考虑铁、铜两种金属，采用成本分块的思想，构建了利润最大化的非线性规划模型，将粒子群算法与神经网络嵌套构成粒子群-神经模型(PSO-ANN)来优化截止品位α_j和入选品位α_r。具体操作为：用截止品位与入选品位一起组成进化计算的粒子群个体，用自适应神经网络建立收益(适应度函数)与各粒子的局部联系，然后利用粒子群算法的全局搜索功能找出使适应度函数(收益)最大时的品位组合。2008年考核计划的可采地质储量为153.17万吨，其地质品位为TFe52.60％，Cu0.306％，其最优截止品位和入选品位分别为14.6848％，入选品位为42.1388％。
(4)兼顾经济效益和资源利用效益，建立截止品位和入选品位的多目标优化模型。首先采用神经网络建立以截止品位和入选品位为变量，精矿量、净现值和资源利用率为目标的函数，再对其方案进行模糊综合评价，将得到的模糊隶属度加权值作为遗传算法的适应度函数，全局搜索出使适应度函数最大，即最优的品位指标组合，实现截止品位和入选品位的动态优化，可为矿山企业提供更为科学合理的决策。
(5)将整个采选系统划分为三个阶段，并建立关键指标(矿量、品位)之间的函数关系式，根据确定的截止品位和入选品位，分别计算各个阶段的矿量和品位，对生产过程的各个环节进行控制和管理。采用截止品位为17．83％，入选品位为46．4％来指导2008年的采选矿生产时，①其采矿阶段，全年全矿损失率为18．31％，贫化率为22．17％，采山矿量为125．123万吨，混岩量35．6415万吨，毛矿量为160．7645万吨，毛矿品位为40．94％，可以通过毛矿量等指标来控制放矿，指导采矿生产；②在配矿阶段，要求入选品位在46．4％左右，假定抛废过程中自产矿混岩量全部抛出，外购矿品位为35％，则需外购矿量80．8278万吨；③选矿阶段全年入选矿量为241．5923万吨，选比在1．9877左右，全年总采出精矿量121．5436万吨，精矿品位在64．5％左右。
Every period of mining and milling in mine system should control the grades of ore. Thegrades of ore contain geological grade, the boundary grade, cut-off grade, grade of crude ore,concentrate grade and so on, and they relate to geological exploration, mining, milling and manyother departments, involving many factors, such as the geological conditions, mining technicalconditions and technical and economic conditions of milling and smelting. Many experts andscholars continue to explore the scientific method for decision-making, the main objectives are tomake full use of resources and to maximize the revenue, which need to find out the optimalcut-off grade and grade of crude ore. Cut-off grade is the grade of ore in the last time of oredrawing during sublevel caving with no sill pillar. If the cut-off grade is low, the dilution rate ofore will be high. This will increase the cost of ore processing, and will decrease the amount of theconcentrate in the condition of a certain mineral processing ability. Contrarily, high cut-off gradewill not only lead to waste of mine resource, but also increase the cost of fundamentalconstruction. Grade of crude ore plays a role as a link between mining and milling. It isdetermined by the average grade of ore bodies, cut-off grade and dilution rate. And it affects themilling recovery and concentrate grade. The optimum grade of crude ore can improve the benefitof mine and make use of mine resource. Therefore, reasonable determination of cut-off grade andgrade of crude ore is crucial to the economic benefit of enterprise and sustainable utilization ofresource, and it has important theoretical and practical meaning.
Because of the particularity of production environment, multiplicity of ore drawing andextension of mine management, in mine enterprise with sublevel caving with no sill pillar,generally, cut-off grade and grade of crude ore are determined by experiment data or worker'sexperience. Along with the increase of resources' scarcity, enhancement of technical andmanagement level, the experts and managers gradually realize that, the determining methodbased on workers' experience, easy and feasible though, greatly increases the mining and millingcost, and wastes resource seriously. With the difference in geological grade and the improvementin production technology, management technology and market economic condition, it is verynecessary to set up mathematical models and relative systems from cut-off grade and grade ofcrude ore to economic benefit and utilization benefit of resource, in order to dynamicallyoptimize cut-off grade and grade of crude ore.
This dissertation is supported by the Project "Research on the optimization of cut-off gradeand production dynamic management in metal mine", which originated from Daye Iron Mine inWuhan Iron and Steel Group Co. Ltd. It mainly resolves two questions. The first one is that, howto determine the cut-off grade for drawing and grade of crude ore, which meet the givenconcentrate grade, not only make the revenue of mining and milling maximum, but also take intoaccount the sustainable utilization of resource? The second one is that, when the optimal cut-offgrade and grade of crude ore are given, how to dynamically control the amounts and grades ofmining ore, milling ore and concentrate ore? So this research aims to resolve the problem ofoptimization and management of cut-off grade and crude ore in metal mine, to provide thescientific method and means for solving the series of questions, in order to upgrade the utilizationof mineral resources, and improve enterprises' economic benefit.
This dissertation mainly researches on the theories and methods of intelligent optimizing twokey grades, namely cut-off grade and grade of crude ore in mining and milling system, byartificial neural network(ANN), fuzzy systems(FS), evolutionary computation(EA) and so on,and take Daye Iron Mine as case studies. Its main contents are as follows.
(1) Propose a novel method that uses numerical simulation to acquire the value of the cut-offgrade in a certain period, use particle swann optimization for structure identification andparameter estimation of the function of loss rate, then construct the function relationship ofcut-off grade and loss rate. Due to the lack of record data of cut-off gradeα_j in practicalproduction, we can't acquire expressionφ(α_j).We know from the apriori information: Thecut-off grade value is normally distributed in certain universe of discourse; the greater cut-offgrade is, the greater loss rate will be, and vice versa. It can be concluded that the distributionfunction of cut-off grade value is f(α_j)= 1/(2π)e~((a_j-18)~2)/2 , where, the mean is u = 18, variance isσ= 1. The numerical simulation algorithm of cut-off grade is proposed, the convergence proofis given, and we successfully obtain cut-off grade values from January 2005 to November 2007in Daye Iron Mine. The Particle Swarm Optimization is used for structure identification andparameter estimation of math model between loss rate and cut-off grade, and the loss ratefunction isφ=1.6508α_j -0.1175, which is the basis of optimizing cut-off grade and grade ofcrude ore.
(2) Evolutionary algorithm and neural network are nested to be a EA-ANN integrated model,and use it to optimize the cut-off grade and crude ore grade. Optimal cut-off grade and grade ofcrude ore are based on maximizing the net present value (NPV). Considering the relationshipamong profit, cost, geological reserves and all kinds of grades, we establish a nonlinear model.The main four parts of intelligent optimization are as follows. The first one is data simulation,which is used to determine relationship between cut-off gradeα_j and loss rateφ; the second isBP neural network, which is to determine relationship between the milling recovery rateεandcut-off grade a_j, crude ore gradeα_r, geological gradeα_t , geological reserves q_t ; the third one is fuzzy system, which is used for determining relationship between the cost of mining & milling Cand cut-off gradeα_j, crude ore gradeα_r, geological gradeα_t, geological reserves q_t,; the lastone is the integration of EA and ANN. The result shows that, during the period of January toNovember in the year 2007, the optimal cut-off grade is 17.8337-17.8367%, and optimal grade ofcrude ore is 46.4%. Comparing with the present scheme (cut-off grade is 18%, grade of crude oreis 41-42%), the optimized scheme can increase the amount of concentrate by 139200 tons, andimprove the net present value by 6.698 million Yuan. Compared PSO-ANN with the GA-ANNand SA-ANN algorithms, it demonstrates that the performance of PSO-ANN is obviouslysuperior to GA-ANN and SA-ANN. A simple input-output interface of grades optimization isdesigned.
(3) Take account into the iron and copper, put total cost into several parts, construct thenonlinear optimization model to maximize the profit. An integration of PSO and ANNs to be aPSO-ANN integrated model, to optimize the cut-off gradeα_j and crude ore gradeα_r. Thedetails are as follows: Join cut-off grade and grade of crude ore together as particle of swarm forevolutionary computation; Make self-adaptive neural network to get the local connectionbetween the income value (fitness function) and each particle; Use PSO algorithm globally tosearch the optimal cut-off grade and grade of crude ore to maximize fitness function. Theproduction plan of Daye Mine in the year 2008 is that, geological reserves is 1.5317 million ton,geological grades are 52.60% and 0.306% for Fe and Cu respectively, through calculation, theoptimal cut-off grade and optimal grade of crude ore are 14.6848%, and 42.1388%, respectively.
(4) Balancing between economic benefit and resource utilization benefit, set up multi-objectiveoptimization model of cut-off grade and grade of crude ore. Neural networks are used toconstruct mapping relationship from cut-off grade, grade of crude ore to the amount ofconcentrate, total present value and total utilization rate; then make fuzzy comprehensiveevaluation of grade combinations, put the weighted fuzzy membership value as fitness functionof genetic algorithm, finally, globally search the best grades combination that make the fitnessfunction maximum, to research dynamic optimization of cut-off grade and grade of crude ore, toassist decision-making and management for mine enterprise.
(5) Divide mining and milling system into three stages, and establish functions of key indexes(amount and grade of ore). Given a certain combination of cut-off grade and grade of crude ore,we can calculate the amount of ore and the grade of ore of every stage, then we can control andmanagement each process of production. If we adopt cut-off grade 17.83%, grade of crude ore46.4% to lead practical production in the year 2008, it can be concluded as follows. In the miningstage, the average loss rate of the whole year is 18.31%, the dilute rate is 22.17%, the amount ofore recovery is 1.25123 million ton, the amount of mixed rock is 356.415 thousand ton, the totalamount and average grade of ore mining are 1.607645 million ton and 40.94%, respectively. Inthe mixing stage, if the grade of ore that should buy from outside is 35%, the amount of ore that need buy is 808.278 thousand ton. In the milling stage, the amount of ore milling is 2.415923million ton, the proportion of the amount of milling ore and concentrate ore is around 1.9877, theamount of concentrate of the whole year is 1.215436 million ton, and the grade of concentrate isaround 64.5%.
引文
[1] 徐忠义，杜前进．采矿知识问答[M]，北京：冶金工业出版社，2006
[2] 杨才亮，马建军．不同放矿截止品位间关系的研究[J]．矿业研究与开发，2005，25(3)，15-18
[3] 黄光球，杜中岳．截止品位和入选原矿品位动态优化方法[J]．江博有色金属，1996，10(4)： 1-5
[4] 黄光球．地下开采矿体边界品位动态优化法[J]．化工矿山技术，1997，26(4)：1-6
[5] 黄光球，桂中岳．地下开采时矿体最小工业品位动态优化的神经网络方法[J]．中国钼业，1997，21(2)：54-58
[6] 余伟健，刘爱辉，明世祥等．金属矿山入选品位动态优化研究[J]．中国钨业，2004，19(3)：27-30
[7] 姜谙男，赵德孝，孙豁然等．进化神经网络在矿山入选品位优化中的应用[J]．矿业研究与开发，2004，24(4)：44-46
[8] 马从安．组合截止品位放矿制度的技术经济分析与决策支持系统[D]．北京科技大学博士论文，2000
[9] 陈少兵．数据包络分析法在矿床品位指标优化中的应用[J]．矿产与地质，1993，7(5)：350-353
[10] 田家华．岩金矿品位指标优化研究的若干问题[J]．矿山地质，1993，14(2)：123-129
[11] 任天贵，高永涛，陈卫星等．论无底柱分段崩落法的截止品位组合[J]．有色金属，1992 (4)： 6-10
[12] 魏一鸣，童光煦．基于神经网络的品位估值方法及其计算机系统[J]．化工矿物与加工，1995，24(2)：15-17
[13] 黄光球．矿石品位推断的神经网络方法[J]．有色冶金设计与研究，1996，17(2)：7-12
[14] 韩万林，张幼蒂．遗传算法在矿石品位估值中的应用[J]．化工矿物与加工，1999(12)：10-13
[15] Xie Yingliang. Optimization of cut-off grade in open-pit based on control theory[J].Transactions of Nonferrous Metals Society of China (English Edition), 1998, 8(2): 353-356
[16] Wooller, R. Cut-off grades beyond the mine Optimizing mill throughput[J]. Monograph Series - Australasian, Institute of Mining and Metallurgy, 2001, 23:459-468
[17] M. Osanloo, M. Ataei. Using equivalent grade factors to find the optimum cut-off grades of multiple metal deposits[J]. Minerals Engineering, 2002, 16:771-776
[18] Poniewierski.J, MacSporran.G, Sheppard. I. Optimization of Cut-Off Grade at Mount Isa Mines Limited's Enterprise Mine[C]. Twelfth International Symposium on Mine Planning and Equipment Selection (Kalgoorlie, WA, Australia), 2003, 4:23-25
[19] Robert D. Cairns, Takayoshi.S. The choice of the cutoff grade in mining[J]. Resources Policy, 2003, 29(3): 75-81
[20] Minnitt, R.C.A. Cut-off grade determination for the maximum value of a small wits-type gold mining operation[J]. Journal of The South African Institute of Mining and Metallurgy, 2004, 104(5): 277-283
[21] Mishra.Brijes. Development of a computer model for determination of cut off grade for metalliferous deposits[J]. Journal of Mines, Metals and Fuels, 2006, 54(6): 147-152
[22] Bascetin. A. Determination of optimal cut-off grade policy to optimize NPV using a new approach with optimization factor[J]. Journal of The South African Institute of Mining and Metallurgy, 2007, 107(2): 87-94
[23] Asad, M.W.A. Optimum cut-off grade policy for open pit mining operations through net present value algorithm considering metal price and cost escalation[J]. Engineering Computations (Swansea, Wales), 2007, 24(7): 723-736
[24] 李荣福．浅谈当次放矿收支平衡法确定截止品位的计算[J]．中国矿业，1995，19(2)：41-44
[25] 刘兴国，张志贵．无底柱分段崩落法不贫化放矿及应用[J]．中国矿业，1998，7(1)：38-41
[26] 尹玉宝．“一品两率”与矿山经济效益关系的探讨[J]．矿业快报，2002(10)：12-13
[27] 王彦武，文启虎．杨家坝铁矿放矿截止品位的确定[J]．黑色金属矿山通讯，1993(2)：14-15
[28] 俞盛建．梅山铁矿放矿截止品位的确定及其技术经济技标分析[J]．黑色金属矿山通讯，1991(2)：9-12
[29] 俞盛建．确定合理放矿截止品位的方法的探讨[J]．化工矿山技术，1992，21(4)：33-37
[30] 任天贵，陈卫星．关于放矿截止品位问题的再探讨[J]．金属矿山，1998(5)：8-14
[31] 马从安，任天贵．截止出矿品位随时间变化的矿床开采模型研究方法探讨[J]．中国地质矿产经济，1998(12)：28-31
[32] 马从安，任天贵．放矿截止品位确定方法探讨[J]．中国地质矿产经济，1999(8)：20-23
[33] 马从安，任天贵．放矿随机过程中散体颗粒移动概率密度方程[J]．河北理工学院学报，1999，21(2)：1-5
[34] 任邦生．涅洁铁矿合理截止品位与出矿品位的研究[J]．矿山地质，1991，12(14)：272-280
[35] 宋海燕．矿山截止品位组合优化及其在生产应用中的研究[D]．北京科技大学硕士论文，1989
[36] 苏志宏．截止放矿品位和截止放矿量的确定[J]．矿山技术，1987(3)：1-7
[37] 白顺祥．铜铁共生矿床采场放矿截止品位及最低可采品位的研究[J]．有色金属(矿山部分)，1992，16-20
[38] 朱春久．合理确定放矿截止品位的经济意义[J]．有色矿冶，1991(6)：1-4
[39] 马维富，周宗红．桃冲铁矿提高井下采出品位的实践[J]．矿业研究与开发，2001，21(3)：7-8
[40] 常云龙．有底柱采矿法合理的放矿截比品位的探讨[J]．矿产资源与保护，1998(3)：50-52
[41] 周传波．无底柱分段崩落法崩矿步距及贫损指标的计算模型与实例[J]．有色矿冶，1994(4)：5-8
[42] 王树棋．自然崩落法放矿控制技术与管理[J]．有色金属(矿山部分)，2002，54(4)：5-8
[43] 武精泊．加强截止品位管理提高矿石回收率[J]．梅山科技，1994(2)：43-44
[44] 李元辉，刘兴国．放矿步距与岩石混入量关系的研究[J]．有色矿冶，2000，16(1)：9-12
[45] 胡杏宝．非均匀度与矿石损失贫化关系的实验研究[J]．化工矿物与加工，2002(1)：16-18
[46] 张资江．确定采场出矿截止品位的探讨[J]．矿产资源与保护，1997(2)：9-10
[47] 李昌宁，宁新亭，任凤玉．矿床开采中的损失贫化分析及对策[J]．世界采矿快报，1999(15)：13-17
[48] 李昌宁，任凤玉，徐小荷．我国矿产资源开采中的损失贫化浅析[J]．自然资源学报，2000，15(1)：36-39
[49] 张国建，郭连军．矿石崩落体形态与放矿效果的关系研究[J]．中国矿业，2001，10(5)：31-33
[50] 王喜兵．无底柱分段崩落法在邯邢矿山的应用实践[J]．金属矿山，1999(6)：11-13
[51] 宋卫东，匡忠祥．程潮铁矿过渡分段合理放矿制度的实验研究[J]．矿业研究与开发，2002，22(4)，23-26
[52] 刘兴国，张志贵．无底柱分段崩落法不贫化放矿理论基础(续一)[J]．金属矿山，1995，7(1)： 23-28
[53] 刘兴国，张志贵．无底柱分段崩落法不贫化放矿理论基础(续二)[J]．金属矿山，1995(12)：13-16
[54] 胡杏保．低贫化放矿工艺现状及应用前景[J]．金属矿山，2002(1)：16-18
[55] 张志贵，刘兴国．无底柱分段崩落法低贫化放矿的工业试验[J]．金属矿山，1997(3)：8-11
[56] 钟一文．智能优化方法及应用研究[D]．浙江大学博士论文，2005
[57] 董聪．计算智能的研究与思考[J]．中国科学基金，2000(1)：6-10
[58] 胡守仁等．神经网络导论[M]．长沙：国防科技大学出版社，1993
[59] Martin T. H, Howard B. D, and Mark B, Neural Network Design[M]. PWS Publishing Company, 1996.
[60] 倪志伟， 李锋刚等．智能管理技术与方法[M]．北京：科学出版社，2007
[61] McClelland J L, Rumelhart D E. Explorations in Parallel Distributed Processing[M]. A handbook of Model, Programs and Exercises.Cambridge, MIT Press, 1986
[62] Rumelhart,D.E, J.L.McClelland. Parallel Distributed processing[M]. Explorations in the Microstrucmre of Cognition: Foundations. MIT press,Cambridge, MA, 1986
[63] Rumelhart, D E, G.E.Hinton, R.J.Williams, Learning representations by back-propagating errors[J]. Nature, 1986, 323(9): 533-536
[64] Moody J, Darken C. Learning with localized receptive fields[C]. Proceedings of the 1988 Connectionist Models Summer School, San Mateo,CA,1988, Carnegie Mellon University, Morgan Kaufmann
[65] Moody J, Darken C. Fast learning in networks of locally-tuned processing units[J]. Neural Computation, 1989, 1(2): 281-294
[66] 韩力群．人工神经网络教程[M]．北京：北京邮电大学出版社，2006
[67] 王立新(王迎军泽)．模糊系统与模糊控制教程[M]．北京：清华大学出版社，2003
[68] E.H.Mamdani and S.Assilian. An experiment in linguistic synthesis with a fuzzy logic controller[J]. International Journal of Man-Machine Studies, 1975, 7(1): 1-13
[69] T.Takagi and M.Sugeno. Fuzzy identification of systems and its applications to modeling and control[J]. IEEE Transactions on Systems, Man and Cysbernetics, 1985, 15:116-132
[70] M.Sugeno and G.T.Kang.Structure identification of fuzzy model.Fuzzy sets and systems,1988, 28:15-33
[71] Y.Tsukamoto. An approach to fuzzy reasoning method[M]. In Madan M. Gupta, Rammohan K. Ragade, and Ronald R. Yager, editors'Advances in fuzzy set theory and applications,North-Holland, Amsterdam, 1979:137-149
[72]张文修，梁广锡．模糊控制与系统[M]．西安：西安交通大学出版社，1998
[73] Holland J H. Adaptation in natural and artificial systems[M]. Ann Arbor: University of Michigan press, 1975.
[74] 李敏强，蔻纪淞．遗传算法的基本理论与应用[M].北京：科学出版社，2002
[75] 王凌．智能优化方法及其应用[M]．北京：清华大学出版社，2001
[76] 席裕庚，柴天佑，恽为民．遗传算法综述[J]．控制理论与应用，1996，13(6)：697-708
[77] 任平．遗传算法综述[J]．工程数学学报，1999，16(1)：1-8
[78] 姚新，陈国良，徐惠敏等．进化算法研究进展[J]．计算机学报，1995，18(9)：694-706
[79] 张晓，戴冠中，徐乃平．一种新的优化搜索算法-遗传算法[J]．控制理论与应用，1995，12(3)：265-273
[80] L.O. Hall, B. Ozyurt, J.C. Bezdek. Clustering with a Genetically Optimized Approach[J].IEEE Trans. on Evolutionary Computation, 1999, 3(2): 103-112
[81] Zhang Wenxiu, Liang Xiu. The Mathematical Base of Genetic Algorithm[M]. Xi'an: Press of Xi'an JiaoTong University, 2000, 5:157-159
[82] 赵正佳，黄洪钟，王金诺．混合离散变量优化的遗传算法研究[J]．中国机械工程，1999，10(12)， 1375
[83] 郭晓婷，朱岩．基于遗传算法的进化神经网络[J]．清华大学学报(自然科学版)，2000 (10)：116-119
[84] 杨晓红，刘乐善．用遗传算法优化神经网络结构[J]．计算机应用与软件，1997(3)：59-65
[85] 梁化楼，戴贵亮．人工神经网络与遗传算法的结合：进展及展望[J]．电子学报，23(10)，1995:194-200
[86] Liu Mingzhe, Wang Ruili etal. A Genetic-Algorithm-based neural network approach for short-term traffic flow forecasting[C]. Lecture Notes in Computer Science, 2005, 3498: 965-970
[87] Zheng D, Gen M, Cheng R. Multiobjective optimization using genetic algorithm[J]. Engineering Valuation and Cost Analysis, 1999, 2:303-310
[88] Li CaiLian, Jia Yong-Xing, Yue Zhen-Jun. Object-weight forecasting model based on adaptive genetic algorithm and BP neural network[J]. Xi Tong Gong Cheng Yu Dian Zi Ji Shu/Systems Engineering and Electronics, 2005, 27(2): 377-380
[89] Yang Yu, YangYanjun. Application of BP neural network improved by genetic algorithm in log interpretation in Luodai gas field[J]. Tianranqi Gongye/Natural Gas Industry, 25(8):47-49
[90] 张智星，孙春在等．神经-模糊和软计算[M]．西安电子科技大学出版社，2002．2
[91] 李士勇等．模糊控制、神经网络和智能控制论[M]．哈尔滨工业大学出版社，1998
[92] 郭嗣琮，陈刚．信息科学中的软计算方法[M]．沈阳：东北大学出版社，2001
[93] 张颖，刘艳秋．软计算方法[M]．北京：科学出版社，2002
[94] 王小平，曹立明．遗传算法-理论、应用与软件实现[M]．西安：西安交通大学出版社，2002
[95] Eberhart R C, Kennedy J. A new optimizer using particle swarm theory. Proc. On 6~(th) International Symposium on Micromachine and Human Science[C]. Piscataway, NJ: IEEE Service Center,2001 : 81-86
[96] Eberhart R C, Shi Y. Particle swarm optimization: developments, applications and resources:IEEE int. Congress on Evolutionary Computation[C]. Piscataway, NJ: IEEE Service Center,2001 : 81-86.
[97] Kennedy J, Eberhart R C. Particle Swarm Optimization: Proc.IEEE International Conference on Neural Networks[C].Piscataway,NJ :IEEE Service Center, 1995:1942-1948.
[98] 谢晓锋，张文俊．微粒群算法综述[J]．控制与决策，2003，18(2)：129-134
[99] 高尚，杨静宁．群智能算法及其应用[M]．北京：中国水利水电出版社，2006
[100] Shi Yuhui, Eberhart R. A modified particle swarm optimizer [A]. Proc IEEE Int Conf. on Evolutionary Computation[C]. Anchorage, 1998: 69-73.
[101] Kennedy J. The particle swarm: Social adaptation of knowledge [A]. Proc IEEE Int Conf.on Evolutionary Computation[C]. Indianapolis, 1997: 303-308.
[102] Thorndike E L. Animal Intelligence: Empirical Studies[M]. New York: MacMillan, 1911
[103] Bandura A. Social Foundations of Thought and Action: A Social Cognitive Theory [M].New Jersey: Prentice-Hall, 1986.
[104] Eberhart R, Shi Yuhui.Particle swarm optimization: Developments, applications and resources [A]. Proc IEEE Int Conf on Evolutionary Computation[C].Seoul, 2001:81-86.
[105] Shi Yuhui, Eberhart R. Parameter select ion in particle swarm optimization[A]. Proc of the 7th Annual Conf on Evolutionary Programming[C]. Washington DC, 1998, 591-600
[106] Shi Yuhui, Eberhart R. Empirical study of particle swarm optimization[C]. Proc of the Congress on Evolutionary Computation, Washington DC, 1999:1945-1950
[107] Suganthan P N, Particle swarm optimizer with neighborhood operator[C]. Proc of the Congress on Evolutionary Computation, Washington DC, 1999:1958-1962
[108] Clerc M. The swarm and the queen: Towards a deterministic and adaptive particle swarm op timization [C]. Proc of the Congress on Evolutionary Computation,Washington DC, 1999:1951-1957
[109] Shi Yuhui, Eberhart R. Fuzzy adaptive particle swarm optimization[C]. Proc IEEE Int Conf on Evolutionary Computation [C]. Seoul, 2001 : 101 -106
[110] Angeline P J. Using selection to improve particle swarm optimization [C]. Proc IEEE Int Conf on Evolutionary Computation, Anchorage, 1998:84-89
[111] Ozcan E, Mohan C. Particle swarm optimization: Surfing the waves[C]. Proc of the Congress on Evolutionary Computation, Washington DC, 1999:1939-1944
[112] Kennedy J. The behavior of particles[C]. Proc 7th Annual Conf on Evolutionary Programming [C]. Gemany: Springer, 1998:581-591
[113] Metroplis N, Rosenbluth A, Rosenbluth Metal. Equation of state calculations by fast computing machines[J]. Journal of Cherimal Physics, 1953, 21 : 1087-1092
[114] Kirkpatrick.S, Gelatt Jr C D, Vecchi.M.P. Optimization by simulated annealing[J]. Science, 1983, 220:671-680
[115] Lundy M, Mess A. Convergence of an annealing algorithm[J]. Mathematical Programming, 1986, 34:111-124
[116] 汪定伟，王俊伟等．智能优化方法[M]．北京：高等教育出版社，2007
[117] 韩万林，张幼蒂，张作祥．BP网络在矿业中的应用[J]．矿业研究与开发，2000，20(6)：48-50
[118] 黄光球．复杂信息条件下矿石品位推断的神经网络方法[J]．西部探矿工程，1996，8(1)：28-32
[119] 王洪德．基于粗集-神经网络的矿井通风系统可靠性理论与方法研究[D]．辽宁工程技术大学博士论文，2004
[120] 王淑红．基于神经网络的选矿厂预测模型及应用研究[D]．昆明理工大学硕士论文，2002
[121] 王凌云，桂卫华，彭晓波．基于智能融合策略的冰铜品位预测模型[J]．化工自动化及仪表，2007，34(4)：26-29
[122] 李勇．磨矿过程参数软测量与综合优化控制的研究[D]．大连理工大学博士论文，2006
[123] 杜玉晓，吴敏，岑丽辉，桂卫华．铅锌烧结过程的集成建模方法及智能优化算法[J]．小型微型计算机系统，2006，25(8)：1458-1463
[124] 汪和平．遗传算法在低贫化放矿中的应用[J]．金属矿山，2003(5)：15-17
[125] Sanchez S, Criado R, Vega C. A generator of Pseudo-random numbers sequences with avery long period [J]. Mathematical and Computer Modeling, 2005, 42 (7-8): 809-816
[126] Nash J C and Walker-Smith M. Nonlinear Parameter Estimation[M]. Marcel Dekker. Inc.,New York.Basel, 1987
[127] 方开泰，张金廷．非线性回归模型参数估计的一个新算法[J]．应用数学学报，1993，16(3)：366-377
[128] 潘正军，康立山，陈毓屏．演化计算[M]．北京：清华大学出版社，1998
[129] Ratkowsky D A. Nonlinear Regression Modeling-a unified practical approach, Marcel Dekker, 1983
[130] 李克庆，刘宝顺．确定合理入选品位的通用系统[J]．北京科技大学学报，1997，19(5)：425-428
[131] 董藏海．矿石品位指标优化方法在二峰山铁矿的应用[J]．金属矿山，2002(4)：14-17
[132] 余伟建，刘爱辉．金属矿山入选品位动态优化研究[J]．中国钨业，2004，19(3)：27-30
[133] 伊晓雨，王占清．鸡冠山铁矿放矿截止品位的分析[J]．矿业工程，2007，5(5)：21-23
[134] 陶勇，李万春．韦钢铁矿采场出矿品位的优化[J]．金属矿山，1995(7)：23
[135] 赵德孝，柳小波，姜谙男．金山店铁矿入选品位动态优化数学模型建立[J]．金属矿山，2004，4：8-10
[136] 李克庆，袁怀雨，胡永平．歪头山铁矿合理入选品位综合评价研究[J]．金属矿山，1998(12)：22-25
[137] 李克庆，袁怀雨，胡永平．合理入选品位整体动态优化技术及其应用[J]．有色金属矿产与勘查．2000，9(3)：24-27
[138] 袁怀雨，刘宝顺，李克庆．合理入选品位整体动态优化[J]．北京科技大学学报，2002，24(3)：239-242
[139] 李克庆，牛京考等．白云鄂博铁矿磁铁矿石品位指标的优化[J]．北京科技大学学报，2007，29(3)：334-337
[140] 杨晓华，沈珍瑶．智能算法及其在资源环境系统建模中的应用[M]．北京师范大学出版社，2005
[141] 焦李成．神经网络系统理论[M]．西安：西安电子科技大学出版社，1990
[142] 岳超源．决策理论与方法[M]．北京：科学出版社，2003
[143] 陈新海．遗传算法的研究与进展[J]．信息与控制，1994，23(4)：215-222
[2] 杨才亮，马建军．不同放矿截止品位间关系的研究[J]．矿业研究与开发，2005，25(3)，15-18
[3] 黄光球，杜中岳．截止品位和入选原矿品位动态优化方法[J]．江博有色金属，1996，10(4)： 1-5
[4] 黄光球．地下开采矿体边界品位动态优化法[J]．化工矿山技术，1997，26(4)：1-6
[5] 黄光球，桂中岳．地下开采时矿体最小工业品位动态优化的神经网络方法[J]．中国钼业，1997，21(2)：54-58
[6] 余伟健，刘爱辉，明世祥等．金属矿山入选品位动态优化研究[J]．中国钨业，2004，19(3)：27-30
[7] 姜谙男，赵德孝，孙豁然等．进化神经网络在矿山入选品位优化中的应用[J]．矿业研究与开发，2004，24(4)：44-46
[8] 马从安．组合截止品位放矿制度的技术经济分析与决策支持系统[D]．北京科技大学博士论文，2000
[9] 陈少兵．数据包络分析法在矿床品位指标优化中的应用[J]．矿产与地质，1993，7(5)：350-353
[10] 田家华．岩金矿品位指标优化研究的若干问题[J]．矿山地质，1993，14(2)：123-129
[11] 任天贵，高永涛，陈卫星等．论无底柱分段崩落法的截止品位组合[J]．有色金属，1992 (4)： 6-10
[12] 魏一鸣，童光煦．基于神经网络的品位估值方法及其计算机系统[J]．化工矿物与加工，1995，24(2)：15-17
[13] 黄光球．矿石品位推断的神经网络方法[J]．有色冶金设计与研究，1996，17(2)：7-12
[14] 韩万林，张幼蒂．遗传算法在矿石品位估值中的应用[J]．化工矿物与加工，1999(12)：10-13
[15] Xie Yingliang. Optimization of cut-off grade in open-pit based on control theory[J].Transactions of Nonferrous Metals Society of China (English Edition), 1998, 8(2): 353-356
[16] Wooller, R. Cut-off grades beyond the mine Optimizing mill throughput[J]. Monograph Series - Australasian, Institute of Mining and Metallurgy, 2001, 23:459-468
[17] M. Osanloo, M. Ataei. Using equivalent grade factors to find the optimum cut-off grades of multiple metal deposits[J]. Minerals Engineering, 2002, 16:771-776
[18] Poniewierski.J, MacSporran.G, Sheppard. I. Optimization of Cut-Off Grade at Mount Isa Mines Limited's Enterprise Mine[C]. Twelfth International Symposium on Mine Planning and Equipment Selection (Kalgoorlie, WA, Australia), 2003, 4:23-25
[19] Robert D. Cairns, Takayoshi.S. The choice of the cutoff grade in mining[J]. Resources Policy, 2003, 29(3): 75-81
[20] Minnitt, R.C.A. Cut-off grade determination for the maximum value of a small wits-type gold mining operation[J]. Journal of The South African Institute of Mining and Metallurgy, 2004, 104(5): 277-283
[21] Mishra.Brijes. Development of a computer model for determination of cut off grade for metalliferous deposits[J]. Journal of Mines, Metals and Fuels, 2006, 54(6): 147-152
[22] Bascetin. A. Determination of optimal cut-off grade policy to optimize NPV using a new approach with optimization factor[J]. Journal of The South African Institute of Mining and Metallurgy, 2007, 107(2): 87-94
[23] Asad, M.W.A. Optimum cut-off grade policy for open pit mining operations through net present value algorithm considering metal price and cost escalation[J]. Engineering Computations (Swansea, Wales), 2007, 24(7): 723-736
[24] 李荣福．浅谈当次放矿收支平衡法确定截止品位的计算[J]．中国矿业，1995，19(2)：41-44
[25] 刘兴国，张志贵．无底柱分段崩落法不贫化放矿及应用[J]．中国矿业，1998，7(1)：38-41
[26] 尹玉宝．“一品两率”与矿山经济效益关系的探讨[J]．矿业快报，2002(10)：12-13
[27] 王彦武，文启虎．杨家坝铁矿放矿截止品位的确定[J]．黑色金属矿山通讯，1993(2)：14-15
[28] 俞盛建．梅山铁矿放矿截止品位的确定及其技术经济技标分析[J]．黑色金属矿山通讯，1991(2)：9-12
[29] 俞盛建．确定合理放矿截止品位的方法的探讨[J]．化工矿山技术，1992，21(4)：33-37
[30] 任天贵，陈卫星．关于放矿截止品位问题的再探讨[J]．金属矿山，1998(5)：8-14
[31] 马从安，任天贵．截止出矿品位随时间变化的矿床开采模型研究方法探讨[J]．中国地质矿产经济，1998(12)：28-31
[32] 马从安，任天贵．放矿截止品位确定方法探讨[J]．中国地质矿产经济，1999(8)：20-23
[33] 马从安，任天贵．放矿随机过程中散体颗粒移动概率密度方程[J]．河北理工学院学报，1999，21(2)：1-5
[34] 任邦生．涅洁铁矿合理截止品位与出矿品位的研究[J]．矿山地质，1991，12(14)：272-280
[35] 宋海燕．矿山截止品位组合优化及其在生产应用中的研究[D]．北京科技大学硕士论文，1989
[36] 苏志宏．截止放矿品位和截止放矿量的确定[J]．矿山技术，1987(3)：1-7
[37] 白顺祥．铜铁共生矿床采场放矿截止品位及最低可采品位的研究[J]．有色金属(矿山部分)，1992，16-20
[38] 朱春久．合理确定放矿截止品位的经济意义[J]．有色矿冶，1991(6)：1-4
[39] 马维富，周宗红．桃冲铁矿提高井下采出品位的实践[J]．矿业研究与开发，2001，21(3)：7-8
[40] 常云龙．有底柱采矿法合理的放矿截比品位的探讨[J]．矿产资源与保护，1998(3)：50-52
[41] 周传波．无底柱分段崩落法崩矿步距及贫损指标的计算模型与实例[J]．有色矿冶，1994(4)：5-8
[42] 王树棋．自然崩落法放矿控制技术与管理[J]．有色金属(矿山部分)，2002，54(4)：5-8
[43] 武精泊．加强截止品位管理提高矿石回收率[J]．梅山科技，1994(2)：43-44
[44] 李元辉，刘兴国．放矿步距与岩石混入量关系的研究[J]．有色矿冶，2000，16(1)：9-12
[45] 胡杏宝．非均匀度与矿石损失贫化关系的实验研究[J]．化工矿物与加工，2002(1)：16-18
[46] 张资江．确定采场出矿截止品位的探讨[J]．矿产资源与保护，1997(2)：9-10
[47] 李昌宁，宁新亭，任凤玉．矿床开采中的损失贫化分析及对策[J]．世界采矿快报，1999(15)：13-17
[48] 李昌宁，任凤玉，徐小荷．我国矿产资源开采中的损失贫化浅析[J]．自然资源学报，2000，15(1)：36-39
[49] 张国建，郭连军．矿石崩落体形态与放矿效果的关系研究[J]．中国矿业，2001，10(5)：31-33
[50] 王喜兵．无底柱分段崩落法在邯邢矿山的应用实践[J]．金属矿山，1999(6)：11-13
[51] 宋卫东，匡忠祥．程潮铁矿过渡分段合理放矿制度的实验研究[J]．矿业研究与开发，2002，22(4)，23-26
[52] 刘兴国，张志贵．无底柱分段崩落法不贫化放矿理论基础(续一)[J]．金属矿山，1995，7(1)： 23-28
[53] 刘兴国，张志贵．无底柱分段崩落法不贫化放矿理论基础(续二)[J]．金属矿山，1995(12)：13-16
[54] 胡杏保．低贫化放矿工艺现状及应用前景[J]．金属矿山，2002(1)：16-18
[55] 张志贵，刘兴国．无底柱分段崩落法低贫化放矿的工业试验[J]．金属矿山，1997(3)：8-11
[56] 钟一文．智能优化方法及应用研究[D]．浙江大学博士论文，2005
[57] 董聪．计算智能的研究与思考[J]．中国科学基金，2000(1)：6-10
[58] 胡守仁等．神经网络导论[M]．长沙：国防科技大学出版社，1993
[59] Martin T. H, Howard B. D, and Mark B, Neural Network Design[M]. PWS Publishing Company, 1996.
[60] 倪志伟， 李锋刚等．智能管理技术与方法[M]．北京：科学出版社，2007
[61] McClelland J L, Rumelhart D E. Explorations in Parallel Distributed Processing[M]. A handbook of Model, Programs and Exercises.Cambridge, MIT Press, 1986
[62] Rumelhart,D.E, J.L.McClelland. Parallel Distributed processing[M]. Explorations in the Microstrucmre of Cognition: Foundations. MIT press,Cambridge, MA, 1986
[63] Rumelhart, D E, G.E.Hinton, R.J.Williams, Learning representations by back-propagating errors[J]. Nature, 1986, 323(9): 533-536
[64] Moody J, Darken C. Learning with localized receptive fields[C]. Proceedings of the 1988 Connectionist Models Summer School, San Mateo,CA,1988, Carnegie Mellon University, Morgan Kaufmann
[65] Moody J, Darken C. Fast learning in networks of locally-tuned processing units[J]. Neural Computation, 1989, 1(2): 281-294
[66] 韩力群．人工神经网络教程[M]．北京：北京邮电大学出版社，2006
[67] 王立新(王迎军泽)．模糊系统与模糊控制教程[M]．北京：清华大学出版社，2003
[68] E.H.Mamdani and S.Assilian. An experiment in linguistic synthesis with a fuzzy logic controller[J]. International Journal of Man-Machine Studies, 1975, 7(1): 1-13
[69] T.Takagi and M.Sugeno. Fuzzy identification of systems and its applications to modeling and control[J]. IEEE Transactions on Systems, Man and Cysbernetics, 1985, 15:116-132
[70] M.Sugeno and G.T.Kang.Structure identification of fuzzy model.Fuzzy sets and systems,1988, 28:15-33
[71] Y.Tsukamoto. An approach to fuzzy reasoning method[M]. In Madan M. Gupta, Rammohan K. Ragade, and Ronald R. Yager, editors'Advances in fuzzy set theory and applications,North-Holland, Amsterdam, 1979:137-149
[72]张文修，梁广锡．模糊控制与系统[M]．西安：西安交通大学出版社，1998
[73] Holland J H. Adaptation in natural and artificial systems[M]. Ann Arbor: University of Michigan press, 1975.
[74] 李敏强，蔻纪淞．遗传算法的基本理论与应用[M].北京：科学出版社，2002
[75] 王凌．智能优化方法及其应用[M]．北京：清华大学出版社，2001
[76] 席裕庚，柴天佑，恽为民．遗传算法综述[J]．控制理论与应用，1996，13(6)：697-708
[77] 任平．遗传算法综述[J]．工程数学学报，1999，16(1)：1-8
[78] 姚新，陈国良，徐惠敏等．进化算法研究进展[J]．计算机学报，1995，18(9)：694-706
[79] 张晓，戴冠中，徐乃平．一种新的优化搜索算法-遗传算法[J]．控制理论与应用，1995，12(3)：265-273
[80] L.O. Hall, B. Ozyurt, J.C. Bezdek. Clustering with a Genetically Optimized Approach[J].IEEE Trans. on Evolutionary Computation, 1999, 3(2): 103-112
[81] Zhang Wenxiu, Liang Xiu. The Mathematical Base of Genetic Algorithm[M]. Xi'an: Press of Xi'an JiaoTong University, 2000, 5:157-159
[82] 赵正佳，黄洪钟，王金诺．混合离散变量优化的遗传算法研究[J]．中国机械工程，1999，10(12)， 1375
[83] 郭晓婷，朱岩．基于遗传算法的进化神经网络[J]．清华大学学报(自然科学版)，2000 (10)：116-119
[84] 杨晓红，刘乐善．用遗传算法优化神经网络结构[J]．计算机应用与软件，1997(3)：59-65
[85] 梁化楼，戴贵亮．人工神经网络与遗传算法的结合：进展及展望[J]．电子学报，23(10)，1995:194-200
[86] Liu Mingzhe, Wang Ruili etal. A Genetic-Algorithm-based neural network approach for short-term traffic flow forecasting[C]. Lecture Notes in Computer Science, 2005, 3498: 965-970
[87] Zheng D, Gen M, Cheng R. Multiobjective optimization using genetic algorithm[J]. Engineering Valuation and Cost Analysis, 1999, 2:303-310
[88] Li CaiLian, Jia Yong-Xing, Yue Zhen-Jun. Object-weight forecasting model based on adaptive genetic algorithm and BP neural network[J]. Xi Tong Gong Cheng Yu Dian Zi Ji Shu/Systems Engineering and Electronics, 2005, 27(2): 377-380
[89] Yang Yu, YangYanjun. Application of BP neural network improved by genetic algorithm in log interpretation in Luodai gas field[J]. Tianranqi Gongye/Natural Gas Industry, 25(8):47-49
[90] 张智星，孙春在等．神经-模糊和软计算[M]．西安电子科技大学出版社，2002．2
[91] 李士勇等．模糊控制、神经网络和智能控制论[M]．哈尔滨工业大学出版社，1998
[92] 郭嗣琮，陈刚．信息科学中的软计算方法[M]．沈阳：东北大学出版社，2001
[93] 张颖，刘艳秋．软计算方法[M]．北京：科学出版社，2002
[94] 王小平，曹立明．遗传算法-理论、应用与软件实现[M]．西安：西安交通大学出版社，2002
[95] Eberhart R C, Kennedy J. A new optimizer using particle swarm theory. Proc. On 6~(th) International Symposium on Micromachine and Human Science[C]. Piscataway, NJ: IEEE Service Center,2001 : 81-86
[96] Eberhart R C, Shi Y. Particle swarm optimization: developments, applications and resources:IEEE int. Congress on Evolutionary Computation[C]. Piscataway, NJ: IEEE Service Center,2001 : 81-86.
[97] Kennedy J, Eberhart R C. Particle Swarm Optimization: Proc.IEEE International Conference on Neural Networks[C].Piscataway,NJ :IEEE Service Center, 1995:1942-1948.
[98] 谢晓锋，张文俊．微粒群算法综述[J]．控制与决策，2003，18(2)：129-134
[99] 高尚，杨静宁．群智能算法及其应用[M]．北京：中国水利水电出版社，2006
[100] Shi Yuhui, Eberhart R. A modified particle swarm optimizer [A]. Proc IEEE Int Conf. on Evolutionary Computation[C]. Anchorage, 1998: 69-73.
[101] Kennedy J. The particle swarm: Social adaptation of knowledge [A]. Proc IEEE Int Conf.on Evolutionary Computation[C]. Indianapolis, 1997: 303-308.
[102] Thorndike E L. Animal Intelligence: Empirical Studies[M]. New York: MacMillan, 1911
[103] Bandura A. Social Foundations of Thought and Action: A Social Cognitive Theory [M].New Jersey: Prentice-Hall, 1986.
[104] Eberhart R, Shi Yuhui.Particle swarm optimization: Developments, applications and resources [A]. Proc IEEE Int Conf on Evolutionary Computation[C].Seoul, 2001:81-86.
[105] Shi Yuhui, Eberhart R. Parameter select ion in particle swarm optimization[A]. Proc of the 7th Annual Conf on Evolutionary Programming[C]. Washington DC, 1998, 591-600
[106] Shi Yuhui, Eberhart R. Empirical study of particle swarm optimization[C]. Proc of the Congress on Evolutionary Computation, Washington DC, 1999:1945-1950
[107] Suganthan P N, Particle swarm optimizer with neighborhood operator[C]. Proc of the Congress on Evolutionary Computation, Washington DC, 1999:1958-1962
[108] Clerc M. The swarm and the queen: Towards a deterministic and adaptive particle swarm op timization [C]. Proc of the Congress on Evolutionary Computation,Washington DC, 1999:1951-1957
[109] Shi Yuhui, Eberhart R. Fuzzy adaptive particle swarm optimization[C]. Proc IEEE Int Conf on Evolutionary Computation [C]. Seoul, 2001 : 101 -106
[110] Angeline P J. Using selection to improve particle swarm optimization [C]. Proc IEEE Int Conf on Evolutionary Computation, Anchorage, 1998:84-89
[111] Ozcan E, Mohan C. Particle swarm optimization: Surfing the waves[C]. Proc of the Congress on Evolutionary Computation, Washington DC, 1999:1939-1944
[112] Kennedy J. The behavior of particles[C]. Proc 7th Annual Conf on Evolutionary Programming [C]. Gemany: Springer, 1998:581-591
[113] Metroplis N, Rosenbluth A, Rosenbluth Metal. Equation of state calculations by fast computing machines[J]. Journal of Cherimal Physics, 1953, 21 : 1087-1092
[114] Kirkpatrick.S, Gelatt Jr C D, Vecchi.M.P. Optimization by simulated annealing[J]. Science, 1983, 220:671-680
[115] Lundy M, Mess A. Convergence of an annealing algorithm[J]. Mathematical Programming, 1986, 34:111-124
[116] 汪定伟，王俊伟等．智能优化方法[M]．北京：高等教育出版社，2007
[117] 韩万林，张幼蒂，张作祥．BP网络在矿业中的应用[J]．矿业研究与开发，2000，20(6)：48-50
[118] 黄光球．复杂信息条件下矿石品位推断的神经网络方法[J]．西部探矿工程，1996，8(1)：28-32
[119] 王洪德．基于粗集-神经网络的矿井通风系统可靠性理论与方法研究[D]．辽宁工程技术大学博士论文，2004
[120] 王淑红．基于神经网络的选矿厂预测模型及应用研究[D]．昆明理工大学硕士论文，2002
[121] 王凌云，桂卫华，彭晓波．基于智能融合策略的冰铜品位预测模型[J]．化工自动化及仪表，2007，34(4)：26-29
[122] 李勇．磨矿过程参数软测量与综合优化控制的研究[D]．大连理工大学博士论文，2006
[123] 杜玉晓，吴敏，岑丽辉，桂卫华．铅锌烧结过程的集成建模方法及智能优化算法[J]．小型微型计算机系统，2006，25(8)：1458-1463
[124] 汪和平．遗传算法在低贫化放矿中的应用[J]．金属矿山，2003(5)：15-17
[125] Sanchez S, Criado R, Vega C. A generator of Pseudo-random numbers sequences with avery long period [J]. Mathematical and Computer Modeling, 2005, 42 (7-8): 809-816
[126] Nash J C and Walker-Smith M. Nonlinear Parameter Estimation[M]. Marcel Dekker. Inc.,New York.Basel, 1987
[127] 方开泰，张金廷．非线性回归模型参数估计的一个新算法[J]．应用数学学报，1993，16(3)：366-377
[128] 潘正军，康立山，陈毓屏．演化计算[M]．北京：清华大学出版社，1998
[129] Ratkowsky D A. Nonlinear Regression Modeling-a unified practical approach, Marcel Dekker, 1983
[130] 李克庆，刘宝顺．确定合理入选品位的通用系统[J]．北京科技大学学报，1997，19(5)：425-428
[131] 董藏海．矿石品位指标优化方法在二峰山铁矿的应用[J]．金属矿山，2002(4)：14-17
[132] 余伟建，刘爱辉．金属矿山入选品位动态优化研究[J]．中国钨业，2004，19(3)：27-30
[133] 伊晓雨，王占清．鸡冠山铁矿放矿截止品位的分析[J]．矿业工程，2007，5(5)：21-23
[134] 陶勇，李万春．韦钢铁矿采场出矿品位的优化[J]．金属矿山，1995(7)：23
[135] 赵德孝，柳小波，姜谙男．金山店铁矿入选品位动态优化数学模型建立[J]．金属矿山，2004，4：8-10
[136] 李克庆，袁怀雨，胡永平．歪头山铁矿合理入选品位综合评价研究[J]．金属矿山，1998(12)：22-25
[137] 李克庆，袁怀雨，胡永平．合理入选品位整体动态优化技术及其应用[J]．有色金属矿产与勘查．2000，9(3)：24-27
[138] 袁怀雨，刘宝顺，李克庆．合理入选品位整体动态优化[J]．北京科技大学学报，2002，24(3)：239-242
[139] 李克庆，牛京考等．白云鄂博铁矿磁铁矿石品位指标的优化[J]．北京科技大学学报，2007，29(3)：334-337
[140] 杨晓华，沈珍瑶．智能算法及其在资源环境系统建模中的应用[M]．北京师范大学出版社，2005
[141] 焦李成．神经网络系统理论[M]．西安：西安电子科技大学出版社，1990
[142] 岳超源．决策理论与方法[M]．北京：科学出版社，2003
[143] 陈新海．遗传算法的研究与进展[J]．信息与控制，1994，23(4)：215-222
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |