【作者】 何玉辉；

【导师】 刘义伦；

【作者基本信息】 中南大学 ， 机械设计及理论， 2010， 博士

【摘要】 热轧板带材作为铝加工的重要产品,已广泛地应用于建筑、包装和交通运输等领域。近年来,热轧板带生产得到了迅猛的发展。板带热轧过程中的温度变化是直接影响产品尺寸精度、力学性能、轧机负荷分配以及能源消耗的重要因素之一,一直是板带生产和研究中关注的重点。然而由于热轧过程中轧件的温度影响因素众多,变形和温度同时存在且相互影响,生产企业大多只能通过昂贵而且耗时的凭经验进行反复试错的方法来调控轧件温度,效果还不理想。随着计算机技术的发展,数值模拟在板带轧制领域的应用越来越广泛。因此,研究板带多道次轧制过程的热力耦合分析基础理论与关键技术,建立准确描述板带热轧过程的有限元模型,借助于先进的研发工具对产品及其制造过程进行快速设计和分析变得尤为重要。本文对1235铝合金热轧过程中流变行为规律进行研究,建立相应的流变应力人工神经网络预报模型,并将其成功地应用于有限元程序中,为铝合金热轧过程的工程计算和数值模拟奠定了良好的基础；开发可准确描述板带多道次热轧过程、参数化的有限元分析系统,数值模拟分析轧件在多道次热轧全流程中的横断面温度场随时间历程的变化规律,得到了可用于自动化生产控制的板带温降数学模型,为板带多道次热轧技术的发展提供一种研究方法。主要研究内容包括：(1)在Gleeble-1500热力模拟机上进行热模拟实验,对1235铝合金在热变形过程中的流变行为进行研究,揭示其流变行为规律,为该材料高温塑性成形工艺的设计、计算及分析提供了理论基础。(2)计入变形温度、应变速率和应变量对流变应力的影响,采用人工神经网络对1235铝合金的高温流变应力进行了预测,并且研究了网络的结构和数据归一化对预测结果的影响。在神经网络的学习过程中,提出将应变速率值对数化、温度值倒数化和流变应力的双曲正弦值对数化的方法,与直接把参数输入到网络相比,新方法的学习效率和预测精度大大提高。通过多次验证,找到了描述1235铝合金流变应力变化规律的最佳神经网络模型为3-15-15-1,其预测值的均方差最大值为0.94MPa,结果表明：神经网络的预测精度远远高于数学模型的回归方法。(3)从塑性加工过程热传导基本方程入手,将力平衡引入能量守恒方程,把温度场和应力场的求解都建立在当前构形上,推导出了基于U.L (UpdatedLagrange)的弹塑性大变形热力耦合分析有限元公式,并给出了详细的求解流程。(4)接触摩擦模拟是有限元分析结果正确与否的关键,也是有限元计算中的一个难题。将热轧变形区沿轧制方向划分为入口滑动区、粘着区和出口滑动区,提出了三区段混合摩擦机理的接触摩擦力计算模型,采用一个形式上和剪切摩擦理论类似的分段函数,有效地解决了剪切摩擦力在中性点不连续的问题,并使纵、横向摩擦力在进入和离开轧制变形区处均为零。(5)如何将人工神经网络预测出的材料流变应力方便、高效地应用于有限元程序,是有限元计算的又一个难题。在MSC.MARC平台下,定义了1235铝合金的用户材料库,实现了流变应力的人工神经网络预报结果和大型通用有限元软件的无缝连接。(6)通过设置与轧制道次相对应的多个载荷工况、建立多个轧辊和推动刚体并设定与轧件的接触关系,在MSC.Marc中实现轧件的顺利咬入和多道次连续轧制过程。开发了一套板带热轧过程有限元分析自动建模系统,建立了某铝厂1235铝合金板带材11道次连续轧制过程的数值仿真模型,得出轧件同一横截面上心部、中部和表面点从出炉到11道次轧制过程的温度变化曲线。计算结果与现场工业实测值吻合。(7)采用正交试验法,研究了轧制速度、乳化液热交换系数、出炉温度和环境温度对板带温降的影响规律。采用回归分析技术,建立了板带温降随工艺参数变化的数学模型表达式,从而为轧制工艺参数设计与分析、节能优化等提供了理论依据。(8)在实验轧机上进行了铝板带热轧温度测试实验,探讨了不同工艺条件对板带温度的影响规律,并将实验测试结果与仿真计算结果进行比较分析,验证了本文建立的热力耦合模型的准确性。更多还原

【Abstract】 As an important aluminium fabrication product, hot rolled aluminum strip has been used widely in many fields including construction, packaging and transportation et al. In recent years, hot-rolled strip production has developed rapidly. The temperature variation of aluminum strip during hot rolling is one of the most important factors that will affect the size accuracy, mechanics capability of the product, the load distribution of rolling mill and energy consumption directly, which has always been the focus of strip production and research. However, there are many factors that influence the strip temperature in the process of hot rolling, and also, deformation and high temperature co-exist and interact with each other, so most manufacturers regulate rolling temperature only relying upon expensive and time-consuming trial and error approach from experience. But the effect is still not ideal. With the development of computer technology, numerical simulation is applied in strip rolling fields more and more widely. Therefore, the study of basic analysis theory and key technology of thermo-mechanical coupling analysis during multi-pass rolling process, establishing an finite element model which can represent accurately the strip rolling process, designing and analyzing the products and manufacturing process with the aid of advanced development tools have become particularly important.In this paper, through the research on the rules of the rheological behavior during 1235 aluminum alloy hot-rolled process, the corresponding artificial neural network prediction model of flow stress was established and applied to the finite element program successfully; a good foundation was laid to engineer calculations and numerical simulation of aluminum hot rolling process; accurate description of the strip multi-pass hot-rolling process and parameterized finite element analysis system was developed; the variation of the cross-sectional temperature field in the whole multi-pass hot-rolling process was analyzed by numerical simulation method; mathematical model of strip temperature drop, used for automated production control, was acquired. A research method was provided for the development of multi-pass hot-rolling technology. The main contents include:(ⅰ) A thermal simulation experiment was operated to examine the rheological behavior of 1235 aluminum alloy during hot deformation and study on the rules of rheological behavior on Gleeble-1500 thermal simulation tester. A theoretical basis was provided for designing, calculating and analyzing high-temperature-plastic forming process of 1235 aluminum alloy.(ⅱ) In view of deformation temperature, strain rate and strain on flow stress, the back propagation (BP) artificial neural network (ANN) was used to predict the flow stress of 1235 aluminum alloy. The input mode is X= (ε, Inε,1/T) and the output mode is Y= ln[sinh(ασ)]. A revised input parameter method and unification algorithm were proposed in this paper which enable the ANN to predict the flow stress accurately in wide range. It is found that the ANN with 3-15-15-1 is the best architecture for predicting the flow stress of aluminum alloy and the maximum mean square of its predicted value is 0.94MPa, with the research results showing that the accuracy of neural network prediction is much higher than that of the mathematical regression method.(ⅲ) To start with the basic equation of heat conduction of plastic processing, the force balance equation was introduced into energy conservation equation. The finite element formulation of elastic-plastic large deformation thermo-mechanical coupled analysis was deduced by building the solving of temperature and stress fields upon the current configuration, and a detailed solution procedure was included.(ⅳ) Contact friction simulation is the key factor to decide whether finite element analysis results are correct or not and also is a difficult problem of the finite element analysis. In the rolling direction, rolling deformation zone was divided into entrance sliding zone, sliding adhesive zone and export sliding zone. The contact friction force calculation model of three-section mixed friction mechanism was proposed using a segment which is similar to the sub-shear friction theory. The shear friction incontinuity in the neutral point was solved effectively, and the vertical and horizontal friction was made to be zero when entering and leaving the deformation zone.(ⅴ)How to use artificial neural network to predict the material flow stress easily and efficiently is another problem in FEM analysis. On the MSC.MARC platform, the research materials-1235 aluminum alloy user library was defined. The perfect combination of artificial neural network prediction model and large-scale finite element software of flow stress was achieved by this method. (ⅵ)By setting the Multiple load cases correlated with the running pass, and then by creating multiple rollers and impellent rigid-bodies, and setting the contact relation with the rolling, the trip can bite smoothly and the multi-pass continuous rolling process was simulated in the finite element software MSC.Marc. A set of automatic Modeling System of finite element analysis for hot-rolled process is developed. We established an strip rolling 11 consecutive numerical simulation model aluminum alloy 1235 strip to get temperature falling curves at center, medium and surface of the same sect of 1235 aluminum strip during the process from outlet of heating furnace to end of 11th pass rolling. The results showed that the calculate result is consistent with measured result.(ⅶ) Studying the law of temperature fall of strip that is influenced by the rolling speed, emulsion heat transfers coefficient, tapping temperature and environmental temperature by orthogonal test. Establishing a mathematical expression/about/under temperature falling with the change of process parameters by regression analysis, a theoretical basis for process parameters design, analysis and energy optimization was provided.(ⅷ) Hot temperature testing laboratories of aluminum strip were carried out on the experimental rolling mill. Discussing the law of the effects to strip temperature of different process conditions and comparing the experimental results with numerical results, the establishment of the coupled thermo-mechanical model was verified.更多还原