【作者】 马恩杰；

【导师】 柴天佑；

【作者基本信息】 东北大学 ， 控制理论与控制工程， 2007， 博士

【摘要】 赤铁矿选矿通过原料处理、竖炉焙烧、磨矿、磁选、精矿以及尾矿处理工序将赤铁矿矿石中的有用元素与脉石及有害元素进行分离,得到铁精矿作为原料供给钢铁工业。综合生产指标是反映企业产品产量、质量、成本以及消耗的指标,赤铁矿选矿厂的综合生产指标包括月的精矿产量、精矿品位、全选比以及原矿消耗量。选矿厂的计划调度主要任务一是根据综合生产指标的目标值产生生产计划及作业计划,最终形成生产指令用来启停选矿生产设备,调配水电气及原料等资源；二是将综合生产指标层层分解到日综合生产指标：各工序(原料处理、竖炉焙烧、磨矿、磁选、精矿以及尾矿处理)作业班的生产量、精矿品位、精矿产量及全选比。工艺技术部门将日综合生产指标转化各工序的工艺指标,如磨矿工序的磨矿粒度、竖炉焙烧工序的一次溢流回收率,然后转化为选矿过程控制系统设定值,控制系统使输出跟踪设定值,从而将综合生产指标控制在目标值范围内。因此,将综合生产指标分解为日综合生产指标的正确与否对能否将选矿厂综合生产指标控制在目标值范围内具有重要意义。选矿厂综合生产指标分解过程具有多层次、多目标、多约束、时变、非线性等特性。多层次表现在需要将月原矿消耗量层层分解到各工序作业班的生产量,将月的精矿品位、精矿产量、全选比层层分解到日；多目标表现在分解过程须按精矿品位、精矿产量、全选比、原矿消耗量的顺序依次保证在目标值范围内,同时要求精矿库存适量,精矿产量均衡,精矿品位的废次尽量减少,各工序作业班生产量均衡；分解过程受设备能力,库存能力,精矿需求,入选原矿的品位、价格、供应能力等多方面约束；其中设备能力,精矿需求等约束具有时变特性；每日库存状态与库存能力,每日精矿产量与设备能力等之间具有强非线性。因此,难以采用已有的计划调度优化方法求日综合生产指标,只能通过人工凭经验分解与试凑来产生日综合生产指标。但人工分解过程费时保守,难以在满足上述约束的条件下保证综合生产指标在目标值范围内。本文在国家高技术研究发展计划(863计划)“选矿工业过程综合自动化系统研究与开发(2004AA412030)”的支持下,以实现综合生产指标的优化控制为目标,开展了综合生产指标分解的优化方法研究,提出了保证综合生产指标在目标值范围内的确定日综合生产指标的优化方法；设计开发了实现上述优化方法的软件包,并应用该软件包进行了综合生产指标优化分解的实验,实验结果验证了所提方法的有效性。本文主要工作归纳如下：(1)提出的日综合生产指标优化方法包括月原矿消耗量优化方法,日原矿消耗量优化方法及各工序作业班生产量优化方法,该方法首先求得月原矿消耗量的最优值并以此为基础通过金属平衡公式求得精矿品位,精矿产量,全选比作为约束求得日原矿消耗量优化值,并以此为基础通过金属平衡公式求得精矿品位,精矿产量,全选比作为约束求得各工序作业班生产量优化值,并以此为基础采用金属平衡公式求得精矿品位,精矿产量及全选比,从而确定了日综合生产指标。月原矿消耗量优化方法以按重要程度与综合生产指标目标值偏差量最小为目标函数,将综合生产指标目标值范围、精矿成本、各种原矿供应总量作为约束条件,建立了以月原矿消耗量为决策变量的非线性目标规划模型,采用了以递归排序为核心的遗传算法求出月原矿消耗量的最优值。日原矿消耗量优化方法以每日精矿产量与月精矿产量的日平均值偏差以及精矿库存量最小为目标函数,以月原矿消耗量优化值为基础通过金属平衡求出的精矿品位、精矿产量、全选比以及每日设备能力、各种原矿混合后品位、精矿库存及精矿品位的上下限作为约束条件,建立以日原矿消耗量为决策变量的非线性多目标规划模型。采用改进的多目标粒子群算法求出了每日各种原矿消耗量的最优值。各工序作业班生产量优化方法以降低每日精矿品位废次惩罚费用以及各工序前后班生产量变化最小为目标函数,将日原矿消耗量优化值为基础通过金属平衡计算出的精矿品位、精矿产量、全选比以及进入到磁选工序的矿石品位、各种缓冲仓料位、精矿品位的上下限作为约束条件,建立了以各工序作业班生产量为决策变量的非线性多目标规划模型,采用改进的多目标粒子群算法求解得了各工序作业班生产量的优化值。(2)应用Microsoft Visual Studio 2005开发环境,编程语言采用C#和Matlab,后台数据库使用SQL Server 2000,基于Web Service技术采用模块化方法设计开发了实现上述优化方法的软件包。该软件由系统管理、月原矿消耗量优化及金属平衡、日原矿消耗量优化及金属平衡、各工序作业班生产量优化及金属平衡、信息查询、系统维护等模块组成。(3)结合某赤铁矿选矿厂综合生产指标分解过程,将该厂综合生产指标目标值：精矿品位52.66%,精矿产量18.23万吨,全选比2.0049,原矿消耗量36.55万吨应用优化软件包进行了分解,得到当月每天的日综合生产指标,以此进行生产可得综合生产指标精矿品位52.69%,精矿产量18.3386万吨,全选比1.9887倍,原矿消耗量36.47万吨,综合生产指标均处于目标值范围内。与人工分解所得实际综合生产指标相比,精矿品位与目标值的偏差由0.11%降为0.03%,精矿产量与目标值的偏差由0.33万吨降为0.1086万吨,原矿消耗量由超出目标值范围的39.53万吨降为在目标值范围内的36.47万吨,全选比由超出目标值范围的2.131倍变为在目标值范围内的1.9887倍,此外,当月精矿产量均衡性提高16.09%,精矿库存量降低5.86%,精矿品位废次从18次减少为6次。实验结果表明本文方法可以用于确定日综合生产指标的优化参考值。更多还原

【Abstract】 Hematite ore dressing process, a procedure of sequential ore processing comprising of raw material treatment, shaft furnace roasting, ore grinding, magnetic separation, and concentrated ore disposal,is adopted to separate the useful elements from the gangue and other harmful ingredients in the hematite ores, while its product of concentrated iron ore will be supplied as the raw material to the steel production industry. Global production indices reflecting the output, quality, cost and production efficiency include monthly concentrate output, concentrate grade, concentration ratio and ore consumption in hematite ore dressing process.One of the objective of planning and scheduling in the dressing process is to determine the production and operating planning based on the target values of global production indices, subsequently, instruction is also determined to start or stop equipments, assign water and power, and so on.The other is that the global production indices are decomposed hierarchically to determine the daily global production indices including processing capacity, concentrate grade, concentrate output and ratio of operation shift. Technical indices of working procedure are determined by the technical department based on the daily global production indices, such as particle size in grinding process, recovery rate of overflow for shaft furnace roasting process.Set points of process control system are determined according to the technical indices, and outputs track the set points. As a result, global production indices can be controlled to achieve their target values.There are multi-layer, multi-objective, multi-restriction, time varying, and nonlinear characteristic in decomposing process for global production indices in mineral-processing factory. For example, monthly ore consumption is decomposed to the output of operation shift; and concentrate grade, output and ratio are decomposed from month to day. Concentrate grade, output and ratio, and ore input, should be controlled in the target range,,as well as stock of concentrate should be minimum and equilibrium, and so on. Lots of restrictions in decomposing process such as equipments, stock, ore grade and ore cost. There are strong nonlinear characteristic between state and capacity of stock, and between daily concentrate output and production capacity of equipments.It is difficult to optimize global production indices using the conventional approach, so manual decomposing for global production indices is adopted in the mineral processing factory today. The process is time-wasting, and the global production indices can not achieve target values while satisfating the restrictions.Combining the the National 863 High Technology Program Project of "Research and Development of Integrated Automatic System on Ore Concentrate Process (2004AA412030)",the research of optimizing decomposing approach for the optimizing control of the global production indices is developed. The optimizing approach to determine the daily global production indices is proposed, which ensures the actual global production indices are controlled in the target range. The optimizing software is designed and developed. The optimizing decomposing approach is tested by some experiments, the results proves the validity of the approach.The main works are as follows:(1)The optimizing approach for daily global production indices proposed in this dissertation consists of optimizing approach for monthly ore consumption, daily ore consumption and processing capcity of operation shift in each process.At first, optimal values for monthly ore input can be obtained with the optimizing approach. Based on optimal values for monthly ore consumption, concentrate grade and output and ratio are determined with the metal equilibrium formula. Considering the restriction for concentrate grade and output and ratio,the optimal value of daily ore consumption can be obtained.Based on the daily ore consumption, concentrate grade and output and ratio are determined with the metal equilibrium formula. Considering the restriction for concentrate grade, output and ratio,the optimal output for operation shift in each process can be obtained. Based on the optimal output for operation shift in each process, concentrate grade, output and ratio are determined with the metal equilibrium formula. After above steps, daily global production indices are determined.Nonlinear objective programming model which decision value is monthly ore consumption is built in the optimizing algorithm for monthly ore consumption, which takes the least error of global production indices based on importance as objective function, and considers the restriction for the target range of global production indices, concentrate cost, the summation of ore supply. Using the improved genetic algorithm as recursion, the optimal monthly ore consumption is determined.Nonlinear multi-objective programming model whose decision value is daily ore consumption is built in the optimizing algorithm for daily ore input, which takes the least error between daily concentrate output and the mean daily concentrate output of one month as objective function, and considers the restriction for the target range of global production indices, concentrate cost, the summation of ore supply. Using the improved particle swarm optimization algorithm,the optimal daily ore consumption is determined.Nonlinear multi-objective programming model whose decision value is the processing capacity of operation shift is built in the optimizing algorithm for operation shift, which takes the least prcessing capacity variety between the previous operation shift and next as objective function, and considers the restriction for the range of the concentrate cost, the summation of ore supply. Using the improved particle swarm optimization algorithm with multi-objective, the optimal processing capacity of operation shift is determined.(2) Base on the Web Service technology and Microsoft Visual Studio 2005,the optimizing software package is build with Visual C# and Matlab programming language and SQL Server 2000 database. The above optimizing approach is implemented using this software package.This software package includes system management module, optimizing for monthly ore consumption and metal equilibrium computation module, optimizing for daily ore consumption and metal equilibrium computation module, optimizing for processing capacity of each operation shift and metal equilibrium computation module, information query module, system maintenance module, and so on.(3)In a certain hematite ore dressing factory, the target values of global production indices are as follows:concentrate grade is 52.66%, concentrate output is 182.3 thousands ton, concentrate ratio is 2.0049, ore consumption is 365.5 thousands ton. Using the optimizing software package, the decomposing results are as follows:concentrate grade is 52.69%, concentrate output is 183.386 thousands ton, concentrate ratio is 1.9887, ore consumption is 364.7 thousands ton, the global production indices are in their target values bound. Comparing with the manual operation, the error between actual and target value of concentrate grade descends from 0.11% to 0.03%,the error between actual and target value of concentrate output descends from 3.3 thousands ton to 1.086 thousands ton, actual value of ore consumption descends from 395.3 thousands ton beyond target range to 364.7 thousands ton in the target range, actual value of concentrate ratio descends from 2.131 beyond target range to 1.9887 in the target range, equilibrium for concentrate output heighten 16.09%,5.86% reduction for concentrate stock is achieved, waster of concentrate grade descends from 18 to 6.The results of experiment show that this approach can provide optimizing reference values of daily global production.更多还原