混合多目标遗传算法及其在热挤压模具型腔形状优化设计中的应用

将复合形法引入遗传算法来反映决策者对多目标问题中各目标函数的偏好信息，提出一种新的结合复合形法的混合多目标遗传算法。针对热挤压模具型腔轮廓形状优化问题，结合刚-粘塑性有限元模拟和神经网络技术，利用三次样条函数插值来表达凹模型腔轮廓形状，以表面载荷沿凹模型腔轮廓表面均匀分布和挤压力最小为目标，建立了多目标优化的数学模型，对挤压模具型腔轮廓形状进行多目标优化设计，得到了最优的凹模形状。对几种不同模具凹模型腔采用MARC／AutoForge有限元软件进行数值模拟对比研究，结果表明，结合复合形法的多目标遗传优化算法是一种较好的模具型腔形状多目标优化设计方法，其优化结果是有效的和显著的。研究结果说明，通过优化型腔形状来提高模具寿命的效果十分显著。     

并行遗传算法在模具型腔形状优化设计中的应用

针时挤压模具型腔形状的优化设计，提出了并行微观遗传优化方法，该方法采用多种群的并行微观遗传算法进行优化计算，利用BP神经网络的预测功能获得目标函数值，采用三次样条插值函数表达挤压模具型腔形状。训练BP神经网络模型的导师信号利用刚塑性有限元数值计算获得。以表面栽荷沿凹模型腔轮廓表面均匀分布为目标，建立了优化数学模型，对挤压模具型腔轮廓形状进行了优化设计。采用有限元软件MARC／AutoForge对优化结果进行了有限元仿真，仿真结果验证了优化结果的有效性。     

基于修正Archard磨损理论的挤压模具磨损分析

基于修正的Archard磨损理论,应用有限元数值模拟软件计算模具挤压成形阶段每个节点的瞬时温度、压力和速度场,以此计算锥形和弧形两种挤压模具型腔的磨损。计算结果表明:挤压模具磨损集中在入口,弧形凹模磨损较锥形凹模磨损大,但弧形凹模磨损均匀,模具寿命较高。磨损计算结果与实际情况较吻合,为预测模具寿命和优化模具型腔奠定了基础。     

基于神经网络与遗传算法的关节轴承挤压模具优化

针对实际生产中关节轴承内外圈间隙分布不均问题,以GEW12关节轴承为例,应用ABAQUS软件对关节轴承挤压装配过程进行了数值模拟。采用BP神经网络建立了挤压模具形状与内外圈最大间隙与最小间隙之差的映射关系。以关节轴承内外圈间隙均匀分布为目标,结合遗传算法,提出了一种集数值仿真、人工神经网络和遗传算法为一体的关节轴承挤压模具型腔优化设计方法。实验结果表明,模具型腔经过优化后,轴承内外圈间隙均匀性得到了很大的改善,轴承金属流动速度更加均匀。     

BP神经网络与遗传算法结合下的温挤压模具优化设计

以汽车转向螺杆类杯-杆件的温挤压凹模为例进行模具磨损分析及其寿命预测。以影响温挤压凹模磨损的4个主要因素,即凹模入口处圆角大小、模具初始硬度、模具初始温度、摩擦因子作为工艺参数,并分别选取4个不同水平值,确定四因素四水平的32组温挤压凹模磨损试验方案,通过Deform 3D有限元数值模拟软件进行成形过程的数值模拟。以不同影响因素和对应模具的磨损量为样本训练BP神经网络,建立4个主要因素与凹模磨损量之间的映射关系,以温挤压凹模磨损量为目标函数,通过遗传算法对4个影响因素进行组合优化,使凹模磨损量最小、寿命最长。     

基于BP神经网络与遗传算法的温挤压模具优化设计

以汽车转向螺杆类杯-杆件的温挤压凹模为例进行模具磨损分析及其寿命预测。以影响温挤压凹模磨损的4个主要因素,即凹模入口处圆角大小、模具初始硬度、模具初始温度、摩擦因子作为工艺参数,并分别选取4个不同水平值,确定四因素四水平的32组温挤压凹模磨损试验方案,通过Deform-3D有限元数值模拟软件进行成形过程的数值模拟。以不同影响因素和对应模具的磨损量为样本训练BP神经网络,建立4个主要因素与凹模磨损量之间的映射关系,以温挤压凹模磨损量为目标函数,通过遗传算法对4个影响因素进行组合优化,使凹模磨损量最小、寿命最长。     

外星轮热挤压预成形多目标优化设计

针对外星轮热挤压成形性能波动,模具杆部过渡圆角区磨损严重的问题,对关键模具结构和挤压工艺参数开展优化研究。基于有限元数值模拟仿真,结合析因试验分析首先得到影响磨损和填充的关键参数及其影响方向。随后对核心参数构造代理模型替代有限元机理模型,用于磨损和填充性能的数值预测。并使用线性加权和法将所得近似模型转化为单目标函数,耦合遗传算法进行全局寻优。计算发现,当入口斜度y=30°,根部圆角a=18.6 mm,杆部圆角b=5 mm,模具硬度HRC60,摩擦因数m=0.1,挤压速度v=10 mm/s,模具预热温度T=300℃,坯料初始温度t=1 180℃时,在保证锻件填充完全的前提下可获得模具最大磨损量的显著降低,模具寿命较初始提高约2 100件。最后通过仿真和生产试验验证了优化结果的正确性。     

扁挤压筒过渡曲面建模

为改善铝型材挤压过程中金属流动状况,通过引入流函数法建立基于流曲线的扁挤压筒型腔到模口之间过渡曲面的边界条件;同时采用保角映射方法,描述扁挤压筒型腔及模口处轮廓分别向单位圆的映射关系,构造过渡曲面入口和出口复杂图形的解析函数,解决了复杂三维模型向准三维对称模型的转换问题。两者结合得到过渡曲面上各点的流线方程,完成对过渡曲面的数学描述。为验证新构建曲面的合理性,对一矩形壁板成形过程进行模拟。结果表明：流线型过渡曲面有利于金属流动,并且有利于降低型腔面上的法向压力,可减少磨损、提高模具寿命。     

基于响应曲面法和粒子群算法的铝型材挤压模具优化设计

导流孔形状是导流模具设计的主要参数，对挤压产品的质量、模具寿命和挤压力有着非常重要的影响。本文以模具出口型材截面的速度均方差为目标，以导流孔的形状为设计变量，采用基于ALE算法的HYPERXTRUDE软件，结合响应曲面法和粒子群算法建立了导流孔形状优化模型并对其进行非线性优化。与优化前的结果相比，优化后型材截面上速度分...     

基于响应曲面法和粒子群算法的铝型材挤压模具优化设计

导流孔形状是导流模具设计的主要参数，对挤压产品的质量、模具寿命和挤压力有着非常重要的影响。本文以模具出口型材截面的速度均方差为目标，以导流孔的形状为设计变量，采用基于ALE算法的HYPERXTRUDE软件，结合响应曲面法和粒子群算法建立了导流孔形状优化模型并对其进行非线性优化。与优化前的结果相比，优化后型材截面上速度分布更加均匀，型材质量得到了大幅度提高。     

Multiobjective optimization design of porthole extrusion die using Pareto-based genetic algorithm

The extrusion die plays a crucial role in aluminum alloy profile production, which influences product quality and service life of extrusion die directly. In this paper, a profile with irregular shape was taken as an analysis example, and multiobjective optimization for porthole extrusion die based on modern intelligence algorithm was carried out. Aiming at achieving the uniform velocity distribution in the cross-section of the profile as well as decreasing the maximum stress on the extrusion die and the deflection of the mandrel, the angle between port bridges, the position of die orifice, and the height of welding chamber were considered as the design variables. Then Kriging model was established on the basis of Latin hypercube samplings, and above design variables were optimized using Pareto-based genetic algorithm. Finally, an optimal die structure is gained. Compared with the initial scheme, the velocity distribution in the extrudate was more even, and the stress on the die and the deflection of the mandrel were decreased obviously in the optimal scheme. The optimal design method for porthole die has strong commonality, thus, it could give useful guidelines for practical production of the same kind of aluminum profile.     

基于多目标优化的扁挤压筒结构设计

在组合式扁挤压筒的结构尺寸设计中，为确保挤压筒最佳工作性能的同时，最大程度地减少过盈装配后内腔产生的变形，提出了多目标优化的概念。结合有限元模拟技术和BP神经网络方法，建立了变过盈量下三层组合式扁挤压筒结构尺寸与各层等效应力分布、内腔位移之间的非线性映射模型，采用多目标遗传算法对其进行优化。优化时，采用了向量评价法、最佳个体保存策略和小生境技术，得到了均匀分布的Pareto最优解，根据定义的满意度函数，选出了最终的满意解。结果表明，在该满意解下，扁挤压筒既实现了等强度设计，又保证了内腔的尺寸精度。     

型材挤压过程工艺参数优化模型

提出了一种集数值仿真、人工神经网络和遗传算法为一体的工艺参数优化模型,用于型材挤压成形过程工艺参数优化,合理配置了非对称角铝型材模模孔位置。通过现场试验,验证了提出的工艺参数优化模型是行之有效和正确的。并在模孔位置优化配置结果基础上对其挤压成形过程进行数值仿真,分析了挤压成形过程中各阶段网格畸变情况,给出了挤压变形时应力和应变分布,对指导型材挤压工艺和模具优化设计具有重要意义。     

数值模拟方法在挤压工艺参数优化中的应用

以出口速度场相对标准速度场偏差(RSDV),平均挤压力P和最大von Mises等效应力σmax作为优化设计目标函数,通过正交设计试验方法和数值模拟手段,对一款双T形截面型材的挤压工艺参数进行优化设计,探讨了一种挤压行业工艺和模具尺寸标准化的方法,并且通过实验验证,证明了经过优化后的工艺合理,实验结果与数值模拟分析结果相吻合。分析发现,在保证模具强度和模具寿命的前提下,提高挤压的速度和模具工作带的高度,能够提高型材挤出的稳定性。     

Die optimization design and experimental study of a large wallboard aluminum alloy profile used for high-speed train

The extrusion die plays a crucial role in the quality control of aluminum alloy profile production. But in practice, the design of extrusion die is mainly dependent on the experience and intuition of die designers; thus, many times of modifications and experiments should be undergone until an acceptable product is gained. In this paper, the extrusion process of a large wallboard aluminum alloy profile used for high-speed train was simulated by means of HyperXtrude software, and the flow behavior of material and deformation mechanism in the die cavity were investigated. With the simulation results of the initial die design scheme, a nonuniform velocity distribution in cross-section of the extrudate was observed. For optimizing the die design scheme, two optimal schemes (adoption of double-step welding chamber and introduction of baffle plate) were proposed. Through optimization, the velocity differences in the extrudate for optimal schemes are decreased from 39.9 to 12.2 and 10.8 mm/s, respectively. Thus, the uniformity of velocity distribution was improved in optimal schemes. The extrusion die design methods for large wallboard profiles were summarized and proposed, including the design methods of baffle plate and double-step welding chamber. Through trial production, a sound wallboard aluminum profile with good geometric shape and high dimensional accuracy was gained. Additionally, the mechanical properties of the extrudate were examined by means of experimental method. It is found that the test results stratified the practical engineering requirements.     

Optimal Shape Design of an Extrusion Die Using Polynomial Networks and Genetic Algorithms

Designing the optimal shape for an extrusion die to produce a high-quality extrusion product is often required by industry. Design from experience is unsatisfactory for achieving the flexibility and precision requirements in die design. In this paper, a design method has been developed for the optimum shape design of extrusion die. The extrusion process was modelled and analysed by using the finite-element method to obtain the extrusion force and effective strain for different die shapes. A polynomial network was applied to identify the force and strain models in terms of the geometric parameters of the extrusion die. An improved genetic algorithm was used to optimise the identified model for optimal shape with minimum force and strain. It has been verified that the modelling error is extremely small. The designer can quickly and accurately access the optimal shape of an extrusion die through this new approach. RID=" ID=" <E5>Correspondence and offprint requests to</E5>: Dr Y.-C. Hsu, Department of Mechanical Manufacturing Engineering, National Huwei Institute of Technology, 64 Wun-Hua Road, Huwei, Yunlin, Taiwan. E-mail: jasonych&commat;sunws.nhit.edt.tw     

基于热力耦合的热挤压模具结构参数优化设计

利用基于有限变形理论的热力耦合基本方程，对方管铝型材平面分流组合模进行了三维热力耦合计算。分析发现应力集中明显，且其部位与模桥裂纹位置相吻合。为改善模具体内应力分布，对该模具的主要结构参数进行了优化设计。经过21次迭代计算，获得最优结构尺寸，使最大应力下降32．2％，从而延长了模具的使用寿命。