【作者】 张立强；

【导师】 李落星；

【作者基本信息】 湖南大学 ， 材料科学与工程， 2011， 博士

【摘要】 为了提高铝合金大型结构件在汽车中的应用,来降低汽车自重,减少油耗和尾气排放,需要探索和开发铝合金新的高效短流程加工工艺。本文在国家科技支撑计划项目“汽车用高性能铝合金整体压铸及高强度钢冲压成形工艺与装备”和湖南省科技重大专项“车辆轻量化关键技术研究与产业化示范”支持下,应用数值模拟技术对薄壁铝合金件金属型低压铸造工艺设计优化及铸件冷却凝固过程中界面换热行为方面做了大量的研究分析工作。本文的工作首先确定铸件冷却凝固过程中的界面换热系数来提高铸造数值模拟的精度,再结合数值模拟技术和优化技术对壁厚1.5mm铝合金薄壁件低压铸造工艺进行设计和优化。具体研究成果如下:1)分别使用两种不同的反分析方法,即反热传导方法和神经网络优化方法,结合实验测得的温度数据求解确定了铸件与金属铸型间的界面换热系数在铸件冷却凝固过程中并不是一个常数,而是不断变化的。并且当铸件温度降到液固两相区时换热系数迅速下降,铸件完全凝固后,换热系数较稳定变化不大。其变化规律为铸造数值模拟边界条件的设置提供了有价值的理论参考。2)为提高界面换热系数求解精度,本文分别建立了一维正热传导与反热传导模型,并编制了相应的计算机程序,通过与温度场的解析解比较证实了正热传导模型的准确性及程序计算的可靠性。并研究分析了反热传导模型中各种计算参数如阻尼系数μ、未来时间步长R及正热传导计算时的时间步长△t等对反算求解结果的稳定性及准确性的影响。3)建立了基于等效比热法的反热传导模型,通过靠近界面位置的温度数据计算界面热流或是界面换热系数可以有效地处理凝固过程的潜热释放问题,并通过此模型计算得到了A356铝合金与铜冷却介质间的界面热流和界面换热系数。结果表明界面热流和换热系数是随铸件凝固时间变化的,其变化范围在1200Wm-2K-1和6200Wm-2K-1之间,而且变化过程中因为结晶潜热的释放存在两个峰值。4)在准确获得铸造数值模拟的重要边界条件-界面换热系数的前提下,考虑到薄壁件金属低压铸造过程中液态金属充型困难的特点,本文同时采用LVQ学习向量化神经网络和BP多层前馈神经网络两种模型构造铸造工艺参数与铸件质量表征间的多目标函数关系。并结合数值模拟与正交实验设计方法代替实际实验来获得神经网络的训练样本,以减小实验成本和提高计算效率。进而再用遗传算法对所建立的函数关系中的参数寻优。5)采用优化的工艺参数制备了壁厚1.5mm的完整铸件,通过与工艺参数优化前仿真和实验结果比较,铸件质量明显改善。为铝合金薄壁件金属型低压铸造技术的研究提供了有价值的参考,并为汽车铝合金大型结构件整体金属型铸造的可行性方面具有宝贵的参考价值。更多还原

【Abstract】 In order to promote large-scale structural aluminum alloy components applied in automotive industry and reduce the vehicle’s own weight and fuel consumption and exhaust emissions, it is necessary to explore and develop an advanced forming technology with high efficiency and short process. Under the support of two project, National Science & Technology Pillar Program“Study on Forming Technology and Equipment of Pressure Cast for High Performance Aluminum Alloy and Stamping for High Strength Steel”and Key Project of Science & Technology of Hunan Province“Research and Demonstration of Light-weight Vehicle Key Technology”, some works on LPDC process design and forming technology for thin-walled aluminum alloy casting with permanent mould are done.In this thesis, the interfacial heat transfer coefficient during casting solidification is calculated to improve the simulation accuracy of the casting process. Then the LPDC process parameters of A356 aluminum thin-walled component with 1.5mm in thickness are designed and optimized using a combining numerical simulation and optimization technology method. The research results in this thesis are summarized as following:1) The casting-metal mold IHTC has been identified by using two different inverse analysis methods, inverse heat conduction method and neural network model, based on measured temperatures at the various locations of casting. The results show that the IHTC is not a constant and varies with time during the casting solidification. And the IHTC decreases rapidly down when the temperature of casting is between liquidus and solidus curve. However, under solidus curve it is almost invariant and remains a constant. This provides valuable references for setting accurately the boundary condition in numerical simulation process.2) In order to calculate the IHTC more accurately, the model of forward and inverse heat conduction problems for one-dimension is established. The code for the models is also compiled. The compared results between the analytical and numerical simulation solution confirm that the model and compiled code of heat conduction for one-dimension is stable and accurate. Moreover, the effects of some parameters in the inverse heat conduction model on the stability and accuracy of calculation results are also discussed in this thesis. The parameters include the damping factorμ, future time step R and time step△t in the forward heat conduction calculation.3) The interfacial heat flux and heat transfer coefficient of A356 casting on a water cooled copper chill are successfully obtained by using the inverse heat conduction method based on the measured temperatures closer to the interface in the casting after the‘equivalent specific heat’method is applied to the forward heat conduction calculation. And it was found that the IHTC varies with time during the casting solidification. The values are in the range of approximately 1200-6200Wm-2K-1 and two peak values exist when the casting temperature is above the solidus temperature because of the released latent heat.4) Considering the poor filling-ability of thin-walled casting in LPDC process, an ANN model combining learning vector quantization (LVQ) and back-propagation (BP) algorithm is proposed in this thesis to map the complex relationship between process conditions and quality indexes of LPDC parts based on the accurate boundary condition, IHTC, in numerical simulation. And the orthogonal array and finite-element method are successfully applied to obtain the training samples for the sake of experimental accuracy and cost saving. Then, a genetic algorithm is implemented to optimize the process.5) By applying the optimized parameters, a thin-walled component with 1.5mm in thickness is successfully prepared. And compared with the casting prepared before process parameters optimized, the quality of casting is obviously improved. Some new results provide the useful guidance for the further study on LPDC technology of aluminum alloy thin-walled component with permanent mould and mass-producing thin-walled aluminum alloy casting applied in automotive industry.更多还原