BOF endpoint prediction based on the flame radiation by hybrid SVC and SVR modeling

A new non-contact method for predicting the endpoint carbon content of different kinds of basic oxygen furnace (BOF) vessel sizes is presented in this paper. The proposed model is based on the spectral distribution of the flame at the vessel mouth with support vector machine (SVM). Two models are constructed with SVM in this paper: a SVC model is used to classify the whole blowing phases based on the mechanism analysis, and an ?-SVR model is constructed to map the flame spectrum variation into the state of the molten steel in the iron bath. The simulation results on industrial datasets show that the proposed method can offer an accurate as well as convenient way to improve the performance of BOF endpoint control compared with current methods of endpoint carbon content prediction.     

Basic oxygen furnace steelmaking end-point prediction based on computer vision and general regression neural network

End-point prediction is one of the most difficult problems in basic oxygen furnace (BOF) steelmaking process. To address this problem, some researchers have proposed some methods based on flame image processing and pattern classification. Because of the dynamically changing flame and real-time needs during the blowing process, there are still some issues that need to be solved. We propose a novel method based on accurate and fast multi flame features extraction and general regression neural network (GRNN). Firstly, flame images were acquired, and then the background of each image was removed via color similarity determination algorithm; secondly, color, texture, and boundary features were extracted; the fast and robust boundary and texture features were extracted by using the proposed methods, and these features were tested for their validity to the end-point prediction via comparing them with some other similar methods; finally, the prediction model was built using multi-features and GRNN. The experimental results demonstrated that it is accurate and fast to use the proposed method to the BOF end-point predict.     

基于炉口辐射光谱支持向量机回归的转炉终点碳含量检测

转炉终点碳含量在线检测对实现炼钢终点准确控制,提高钢铁产品质量,降低能耗,减少废气排放具有重要意义。针对转炉冶炼终点控制及碳含量检测的难题,研究了一种新的基于炉口辐射光谱分析的非接触式在线碳含量检测方法。方法基于辐射光谱的支持向量机回归(SVR)实现转炉终点前过程的碳含量预测。通过远距离光谱采集系统获取炉口火焰光谱信息,基于冶炼过程炉口火焰辐射光谱变化规律的分析,分别提取了表征辐射光谱整体特征的两个参数即总谱宽和辐射峰值、以及表征发射谱的三个特征波长600,630和775 nm处的幅值作为支持向量机的输入,结合脱碳理论和实测碳值拟合重构的脱碳函数曲线作为支持向量机的输出,利用支持向量机回归方法建立光谱分布与碳含量的关系模型。通过训练样本集和测试集循环优化确定模型最佳参数。设计的仪器和优选的模型已安装在转炉生产现场长时间运行,现场实验结果表明,终点碳含量检测准确率为90.2%,测量时间小于0.3 s,可实时在线检测,能够满足生产需求,为转炉冶炼终点的精确控制提供了重要依据。     

高炉铁水Si含量的修正混沌加权一阶局部预报

基于混沌加权一阶局部预测模型，在拟合预测器过程中，采用矩阵、向量拟合取代单 一变量拟合，对预测模型进行了修正，并将修正的预测模型应用于国内中型高炉（380—750m^3）Si含量预报，取得了很好的效果.同时得出高炉冶炼过程的混沌特征状态量Kolmo gorov熵值，影响着高炉Si含量预报命中率的高低. 关键词： 高炉冶炼 铁水Si含量预报 混沌加权一阶局部预报 Kolmogorov熵     

WCARS-ISPA火焰光谱特征选择的转炉炼钢终点预测

对转炉炼钢终点的实时精准控制能够有效提高钢铁产出的质量,炉口火焰光谱在炼钢不同时期的变化明显,对其进行分析处理并与机器学习方法相结合可有效用于炼钢终点的实时控制。针对炉口火焰光谱数据量大、现有方法对光谱特征提取在可信度和实时性上不足的缺陷,提出一种基于窗口竞争性自适应重加权采样（WCARS）结合迭代式连续投影算法（ISPA）的光谱特征波长选择方法,该方法在有效解决模型过拟合问题的同时,能够降低高维数据计算的复杂度。将火焰光谱数据沿波长方向进行窗口划分后,使用CARS进行计算选出特征窗口波段,再将迭代式选择与传统连续投影算法相结合,通过重复迭代精选出特征波长,在此基础上使用支持向量机回归（SVR）建立炼钢终点碳含量预测模型。实验采集363组炼钢后期的炉口火焰光谱数据作为样本,并对其进行Savitzky-Golay平滑预处理。使用WCARS-ISPA算法从全光谱数据中选出10个特征波长作为SVR模型的输入,碳含量为模型输出,Kennard-stone算法对训练集和测试集进行划分,选择碳含量的平均预测误差、预测误差在±2%以内的命中率以及运行30次的平均时间作为模型评价指标。实验结果显示,模型的平均碳含量预测误差为1.413 2%,命中率高达90.63%,运行时间为0.019 679 s。与使用全光谱和WCARS-ISPA,CARS-SPA,WCARS和SPA四种不同特征选择方法选出的特征波长建模得到的结果进行对比,基于WCARS-ISPA方法选出的特征波长建立的终点碳含量预测模型误差最小、命中率最高。提出一种新的炉口火焰光谱特征波长提取方法,使用窗口竞争性自适应重加权采样结合迭代式连续投影算法选取特征波长,并在此基础上建立转炉炼钢终点碳含量预测模型,实验结果表明,该方法能够有效提取火焰光谱特征,所建模型能够对转炉炼钢终点进行准确预测,满足工业生产的实时控制要求,为实际生产提供可靠帮助。     

基于傅里叶红外光谱的钢渣代炭黑环保型复合橡胶的补强机理及性能研究

以铁水脱硫渣作为研究对象,利用铁水脱硫渣作为橡胶填料取代部分炭黑与丁苯橡胶进行复合,制备铁水脱硫渣/丁苯橡胶。利用多种方法测试铁水脱硫渣/丁苯橡胶的性能,采用傅里叶变换红外光谱仪测试硫化过程中不同阶段铁水脱硫渣的结构组成。结果表明：利用铁水脱硫渣部分替代炭黑,可达到补强效果与降低补强剂成本的目的。铁水脱硫渣/丁苯橡胶的正硫化时间(t₉₀)为25.08 min,其焦烧期为0~15 min、热硫化期15~25 min和硫化平坦期25~45 min。在焦烧期铁水脱硫渣可以提供碱环境,利于增加丁苯橡胶流动性;在热硫化期与硫化平坦期,铁水脱硫渣中Ca₂SiO₄能够持续加速发生水化反应生成C-S-H凝胶,达到对丁苯橡胶补强的效果。另外,铁水脱硫渣可以避免铁水脱硫渣/丁苯橡胶过硫化期的出现。     

火焰特征发射谱线研究

本文利用CCD光纤光谱仪,通过实验确定了木基燃料燃烧火焰光谱中出现的特征谱线是K,Na 的原子发射谱线,测定了特征谱线出现时的火焰温度,指出可以通过特征谱线的出现与否来确定火焰的温度范围。还发现火焰的光谱形状和火焰的燃烧状态密切相关,包含大量和燃烧有关的信息,并且对煤烟颗粒有较强的抗干扰性。通过对火焰光谱形状的分析可判断出火焰的燃烧状况,进行燃烧诊断。     

Determination of Zn distribution and speciation in basic oxygen furnace sludge by synchrotron radiation induced μ‐XRF and μ‐XANES microspectroscopy

Basic oxygen furnace off gas (BOF OG) sludge is one important type of metallurgical residues in steelmaking industry. The presence of zinc element is one key factor that constrains the practical use of such metallurgical residues. Information about the chemical association of zinc in BOF OG sludge is fundamental for understanding its behavior during the formation and further treatments. In this paper, the spatial distribution and the chemical forms of zinc in accumulated particles of BOF OG sludge have been investigated using synchrotron radiation induced μ‐X‐ray fluorescence and μ‐X‐ray absorption near edge spectroscopy microspectroscopy. Results of μ‐X‐ray fluorescence analysis showed that zinc distributed in two ways. One was shared with iron, and its distribution showed a positive correlation with that of iron. The other was accumulated in some well‐defined hot spots with high amount and its distribution showed negative correlation with that of iron. And results of μ‐X‐ray absorption near edge spectroscopy microspectroscopy indicated that ZnFe₂O₄ was the main constituent in some well‐defined hot spots, whereas in other cases, zinc was mainly in the form of zinc carbonate. Copyright © 2013 John Wiley & Sons, Ltd.     

估测田间烟叶色素含量的光谱模型研究

田间烟叶色素含量的光谱无损快速测量,对烟草营养生长期的营养诊断与长势监测、成熟期的烟叶品质评判具有重要的生产指导意义。该研究的目的是利用烟叶田间光谱估测烟叶的叶绿素和类胡萝卜素含量。研究采集了营养生长期和成熟期烟叶田间反射光谱,测量了样品烟叶的色素含量,利用支持向量机(SVM)和光谱指数法,对营养生长期和成熟期烟叶样品用分期建模和混合建模两种方法建立色素含量估测模型,并对模型的预测性能进行比较。研究结果表明,分期建模和混合建模对于烟叶色素含量的估测效果差异不显著。对于叶绿素含量,SVM和光谱指数法均有较好的估测效果;对于类胡萝卜素含量,SVM方法比光谱指数法具有更高的估测精度。采用SVM方法对烟叶样品的叶绿素含量分期建模得到的估测决定系数和均方根误差分别为0.862 9和0.015 5,对叶绿素含量混合建模得到的估测决定系数和均方根误差分别为0.898 5和0.012 3;采用SVM方法对烟叶样品类胡萝卜素含量分期建模得到的估测决定系数和均方根误差分别为0.873 0和0.002 4,对类胡萝卜素含量混合建模得到的估测决定系数和均方根误差分别为0.852 7和0.002 4。该研究的创新点是通过支持向量机和光谱指数法采用分期建模以及混合建模的方式建立了烟叶样品色素含量的估测模型,为烟草田间生产的质量控制、烟叶的采收品质保证提供科学依据和技术支持。     

Improvement in sonochemical degradation of 4-chlorophenol by combined use of Fenton-like reagents

Studies on the sonolysis of a wide range of organic compounds have demonstrated that ultrasonic irradiation has potential for decomposition of organic pollutants in hazardous wastewater. However, the ultrasonic irradiation alone cannot provide high enough rate of decomposition to be used practically. One of the solutions to increase the degradation efficiency is to combine the ultrasound application with other advanced chemical oxidation processes (AOPs). In this study, in order to increase the efficiency of ultrasonically assisted degradation of organic pollutants in water, we examined effects of three kinds of solid Fe-containing catalysts, namely iron powder, basic oxygen furnace (BOF) slag and mill scale on the degradation rate of 4-CP (4-chlorophenol) in aqueous solutions containing hydrogen peroxide. In the experiments, 4-CP was considered as a model organic compound. All three Fe-containing matters when react with hydrogen peroxide are involved in the Fenton-like reaction system, which is one of the promising AOPs. The results showed that both the iron powder and mill scale additions can accelerate the degradation of 4-CP, although the effect is dependent on the solution pH. All 4-CP could be decomposed for 2 min at pH=3 and for 1h at pH=5.6. On the other hand, the BOF slag had no catalysis effect on the 4-CP degradation because of higher concentration of calcium and lower concentration of iron.     

基于炉口辐射和改进神经网络的转炉终点预测模型

针对国内外转炉炼钢终点控制的现状,建立了一种用于终点预测的神经网络模型.以炉口辐射信息获取系统为实验平台,运用光纤谱分复用和颜色空间模型转换技术.分析发现了光谱与图像信息特征量在吹炼过程中晕现出中前期类似、末期相反的规律.从得到的特征规律曲线中选用一些关键特征量,在改进的修正系数算法基础上,进行了模型的训练和预测分析.实验结果表明:响应时间在2 s以内,满足快速判定的时间要求;改进算法的模型预测精度高于常规算法,该系统可以止常工作在转炉炼钢的恶劣环境下,达到了预期效果.     

高炉冶炼过程的混沌性解析

根据Grassberger和Procaccia提出的Kolmogorov熵计算方法，以山东莱芜钢铁集团公司1号 高炉和山西临汾钢铁集团公司6号高炉测得的铁水硅含量时间序列为样本，计算了各自的Kol mogorov熵分别为（01453±00151）nats·h-1^-1和（01553±0014 0）nats·h-1^-1，并估计了两座高炉铁水硅含量可预测的时间尺度分别约为6 88和644h.由Kolmogorov熵均为大于零的有限值，定量地说明了这两座高炉 关键词： 混沌 Kolmogorov熵 高炉冶炼过程 铁水硅含量     

基于遗传算法优化BP神经网络的高炉喷煤优化

高炉炼铁是一个复杂的多变量系统,而现行的操作制度是基于炉长经验的参数设置模式,导致能源尤其是煤粉的消耗常常处于“盲目”状态。本文综合炼铁工艺理论和高炉专家经验,针对白云鄂博矿石冶炼的特殊性,采用筛选出的优化数据,利用遗传算法所固有的全局搜索性能优化BP神经网络模型的权值和阈值,分别建立了基于遗传算法优化BP神经网络的高炉喷煤量优化预测模型以及工艺指标(铁水[Si]含量及入炉焦比)预测模型。优化数据的利用使得上述模型可以根据高炉当前炉况输出喷煤量的最佳优化设定值,并预测出相对应的工艺指标变化趋势。实际应用表明,本方法能够给现场操作人员提供操作指导,实现高炉稳定顺行、提高经济效益的目的。     

基于LIBS技术的钢铁合金中Cr和Ni元素SVM定量分析方法研究

针对钢铁合金样品中存在基体效应复杂的问题,通过优化支持向量机模型的输入特征,建立多元素变量的定量分析模型,预测钢铁合金样品中Cr和Ni元素的含量。研究结果表明,分别以特征谱线的峰值强度和积分强度作为支持向量机模型的输入时,积分强度因为包含了谱线的谱宽和形状信息,模型训练效果较好;相比于单一元素谱线的特征信息,采用多元素的多条谱线信息输入支持向量机模型时,模型训练效果较好,这是由于多种谱线信息的输入可以有效校正基体效应的影响。在此基础上,通过归一化变量将内标法与多变量定标方法有效结合,不仅可以减小实验测量误差还能有效校正基体效应的影响,而且有效提高了模型的重复率和准确率。归一化变量作为支持向量机模型的输入变量,对待测样品S₁和S₂中Cr元素含量预测的相对误差为6.58%和1.12%,对Ni元素浓度预测的相对误差为13.4%和4.71%。通过归一化变量将内标法与多变量定标方法有效结合,可以充分发挥SVM算法的非线性学习优势,为LIBS技术应用于复杂样品定量定标分析提供理论基础。     

氧气顶炉口火焰温度多光谱分析

采用多波长分析方法测量氧气顶回转炉口火焰温度的研究。在该方法中,采用USB4000型光纤光谱仪测量火焰在可见光波长范围内的火焰辐射光谱,结合Levenberg-Marquardt最优化算法,得到火焰温度和单色辐射率变化规律。本文还提出了采用小波神经网络处理多光谱测温数据,该方法可以取消发射率与波长之间的假设模型,是一种获得目标真温及发射率行之有效的方法。网络中隐含层的神经元是由S函数和Morlet小波函数的乘积组合构成,在逼近火焰温度中具有良好的表现。     

基于支持向量机的中药太赫兹光谱鉴别

文章将支持向量机用于中药材太赫兹光谱识别.利用太赫兹光谱系统测得三组相似中药炙甘草和生甘草、南柴胡和北柴胡、山豆根和北豆根的太赫兹光谱,傅罩叶变换后得到它们的吸收系数作为分类鉴别的特征数据.用线内积函数、多项式内积函数和径向基内积函数分别构建三种小同的支持向量机,并建立误差反传神经网络(BP神经网络),分别用支持向量机和BP神经网络对中药的特征数据进行鉴别.识别结果比较表明,支持向量机在小样本情况下对中药两分类识别的效果明显超过BP神经网络.     

Microscopically derived potential energy surfaces from mostly structural considerations

A simple procedure to estimate the quadrupole Potential-Energy-Surface (PES) is presented, using mainly structural information, namely the content of the shell model space and the Pauli exclusion principle. Further microscopic properties are implicitly contained through the use of results from the Möller and Nix tables or experimental information. A mapping to the geometric potential is performed yielding the PES. The General Collective Model is used in order to obtain an estimate on the spectrum and quadrupole transitions, adjusting only the mass parameter. First, we test the conjecture on known nuclei, deriving the PES and compare them to known data. We will see that the PES approximates very well the structure expected. Having acquired a certain confidence, we predict the PES of several chain of isotopes of heavy and super-heavy nuclei and at the end we investigate the structure of nuclei in the supposed island of stability. One of the main points to show is that simple assumptions can provide already important information on the structure of nuclei outside known regions and that spectra and electromagnetic transitions can be estimated without using involved calculations and assumptions. The procedure does not allow to calculate binding energies. The method presented can be viewed as a starting point for further improvements.     

基于小波分析的ARIMA与LSSVM组合的高炉煤气预测

高炉煤气发生量的准确预测对钢铁企业能源优化调度具有重要意义。针对钢铁企业中基于机理模型的高炉煤气发生量难以准确预测问题,建立了基于小波分析的最小二乘支持向量机(LSSVM)和自回归差分滑动平均(ARIMA)相结合的高炉煤气预测模型。预测前利用小波去噪对原始数据进行消噪处理,并对处理后的数据进行小波变换得到趋势序列和波动序列,然后对各部分序列分别建模和预测,最后将各部分预测结果叠加;仿真结果表明,组合预测模型减小了预测误差,提高了预测精度。与其他模型相比,组合预测模型更适合高炉煤气预测。     

Molecular dynamics simulation of iron nanoparticle sintering during flame synthesis

The sintering process of iron nanoparticles produced in a flame environment is investigated by molecular dynamic (MD) simulations. The thermodynamic characteristics and restructuring pathways are studied for two-body and three-body sintering processes. The melting point, energy, and structures are computed for nanoparticles before and after sintering. A simplified model is proposed to predict the equilibrium temperature of nanoparticles upon sintering. The MD results are used to explain the formation mechanisms of two size ranges of nanoparticles during the flame synthesis. The role of sintering during nanoparticle growth is analyzed.     

基于多分辨分析的高炉铁水含硅量波动多重分形辨识

以包钢6号高炉、邯钢7号高炉和莱钢1号高炉在线采集的铁水含硅量(［Si］)的时间序列为样本, 利用多分辨分析剔除样本的长期趋势,对样本保留的波动趋势进行多重分形特征辨识. 通过计算广义Hurst指数、尺度函数、多重分形谱, 全面、细致量化了序列的局部及不同层次的波动奇异性. 计算结果表明: 去除长期趋势后, 三座高炉［Si］序列的波动呈现显著多重分形特征, 这样的波动过程仅用单一的Hurst指数或box维数来描述是不够的. 关键词： 多分辨分析 铁水含硅量 波动 多重分形