【作者】 梁栋；

【导师】 白晨光；

【作者基本信息】 重庆大学 ， 冶金工程， 2008， 博士

【摘要】 高炉炼铁系统是钢铁冶金过程中的重要部分,整个流程工序繁多,工艺机理复杂,现象解释困难。对关键过程工艺操作建立优化模型,对维系生产稳定,提高生产效率,改进产品质量,降低生产成本具有重要作用。迄今为止,国内外已有许多针对高炉炼铁过程优化的研究,建立了各具特色的数学模型,但是这些研究中仍存在着一些不足,如烧结配料优化中仅着眼于化学成分的约束;高炉布料优化缺乏系统性的研究,模型的一些假设条件与现实难以吻合;高炉炉况判断及预测模型复杂,知识获取困难等。而对于“烧结-高炉”工艺过程的调度优化模型研究也鲜有报道。本文针对高炉炼铁系统中,烧结配料,高炉布料,高炉炉况判断以及区域调度分别建立了工艺预测及优化模型。①烧结配料优化模型中:1)运用遗传算法对重钢烧结配料进行了优化,在保证烧结矿化学成分、物理性能等综合指标符合入炉要求的前提下,降低了成本,增加了烧结矿各项指标的稳定性。2)运用多元逐步回归方法,建立起了烧结矿物理性能与配矿的函数关系。对指导当前配料基础上的优化配矿有良好效果。解决配料过程中烧结矿冶金性能的约束可以用同样的方法。②高炉布料模型中:1)运用传统力学理论,结合炉料布料过程中运动关系、炉料堆积堆角修正O/C比预测等模块建立了高炉无料钟布料模型。其中,对于料流轨迹采用了两条料股的方式,更符合物理规律,对于料面形状的处理采用了多段线性化处理。2)结合遗传算法,建立了基于目标O/C比布料操作优化模型,可较好的实现目标O/C比下的布料参数优化。3)建立1:15的物理模型,通过实验验证了模型主料流轨迹及下料流轨迹;并通过实验验证了操作参数对于不同粒度的原料布料堆尖位置的影响。4)建立了基于图像处理检测固体散料的方法,以此测量了实验布沙形成的堆角,研究了不同操作参量下堆积过程中堆角的变化规律。③高炉炉况判断模型中:1)结合遗传算法以及模糊聚类,建立了基于数据挖掘的炉况判断模型。通过优化可自主选择最佳的建模方法及相应参数;可完全经由数据分析而自行生成基础规则库,打破传统专家系统知识获取的困难;模型简易,程序编写简单,易于维护与进一步的开发。今后将实时数据存库,定期重复上述过程,将会使系统具有自适应进化功能。2)提出以小波分析结合支持向量的技术路线对高炉生产中的透气性指数进行预测,通过与其它人工智能方法建立的预测模型比较,其预测精度高,拟合与推广能力也比较好。④“烧结-高炉”工序调度优化模型中:1)以炼钢需求为订单,使用弧赋时Petri网(TAPN)对“烧结-高炉”工序进行建模,提出了包括关键操作的连续性函数、满足订单的供应函数和关键仓贮的稳定性函数在内的复合目标函数。最小化目标函数为目标通过遗传算法搜索最优的调度参数。2)以某炼铁厂生产数据代入模型,通过仿真优化后的调度参数:可以满足关键操作连续;可以保证最大限度完成订单;可以保持关键仓储最佳的稳定性。3)将模型逆转可实现物流追踪功能。4)模型进行简单的修改,引入检修计划,可以实现检修计划下的调度优化。更多还原

【Abstract】 The procedure in blast furnace ironmaking includes many complicated sub process. It is hard to explain some phenomena clearly. Prediction and optimization for operation of the section can maintain stable process, improve quality and quantity of products and reduce the cost of products.Presently, there are some limits in the known forecasting and optimal model for ironmaking. Most model only emphasis on optimization of processing operation, there are no mature scheduling models reported or used in this section.Aiming at sinter proportioning, burden distribution of blast furnace, judge and prediction of operational state of BF and scheduling in“Sinter-BF”section, forecasting and optimal models are built.①In the optimized model for sinter proportioning:1) GA (Genetic Algorithm) is used in optimizing sinter proportioning. The model can guarantee the chemical and physical demand of sinter and reduce the cost of sinter.2) Multiple stepwise regression is used to build the relation between physical property of sinter and sinter proportioning. It is helpful to optimize sinter proportioning.②In the model for burden distribution of blast furnace:1) Combined Newton law of motion with angle correction, prediction of O/C, a simulation model for burden distribution of BF is built. In the model, two trajectory lines are used as burden flow trajectory, which is accord with reality. The burden profile is used linearization treatment.2) Aiming at minimal difference of O/C between model and object, model can optimize operational parameter accord to object O/C with GA.3) An experimental setup that proportion of size is 1:15 is used to verity the burden trajectory and effect of operational parameters to different size burden.4) Image processing method is used to study change law of dumping angel with experiment.③In the model for intelligent forecasting and judge operational state of BF:1) An intelligent judging model for operational state of BF is built with GA and fuzzy cluster. The model can be built in data mining automatically.2) An intelligent forecasting model for permeability index is built with wavelet analysis and support vector machine. Compared with other intelligent techniques, the model has higher forecasting accuracy and generalized ability.④In the scheduling model:1) The model integrated Petri Net and GA (genetic algorithm) as suitable tools for simulation and optimization of scheduling to sintering and BF process.2) By analysis, the conclusion is drawn that scheduling with optimized values can maintain continuity of key operations , can satisfy order of hot metal in maximal degree and can keep the best stability of key storages.3) The model can easily fulfill logistics traceability with reversing model.4) The model can fulfill optimization scheduling in maintenance plan.更多还原