【作者】 周泽；

【导师】 李光耀；

【作者基本信息】 湖南大学 ， 车辆工程， 2013， 博士

【摘要】 汽车结构的疲劳寿命决定汽车的使用年限，通常采用实车试验或计算机辅助技术来评估其疲劳寿命。实车试验结果真实准确，但试验周期较长且费用巨大；而计算机辅助技术受制于虚拟仿真与实车试验的差异及疲劳理论的缺陷，其预测精度较低，工程应用价值有限。目前，汽车结构的疲劳寿命评估依赖于实车试验，导致汽车产品的研发周期较长、研发费用巨大。本文对计算机辅助技术应用于汽车结构的疲劳寿命预测领域开展系统的研究，提出基于真实路谱重现的虚拟台架方法来提高虚拟仿真的准确性，并应用于汽车结构的疲劳寿命预测中。该方法在汽车产品研发中，能够快速而准确实现汽车结构的疲劳寿命预测，指导结构的疲劳性能提升，从而减少实车试验次数，缩短汽车产品研发周期和降低研发费用，具有重要的理论和工程应用价值。本文的主要内容和创新点包括：（1）提出基于真实路谱重现的虚拟台架方法，提高虚拟仿真模拟汽车实车试验的精度。采用计算机辅助技术进行汽车实车试验的虚拟仿真时，由于路面-轮胎-整车耦合关系复杂，特别是轮胎的模拟，为当前虚拟仿真技术难题，限制了疲劳寿命预测精度提高。本文提出基于真实路谱重现的虚拟台架方法，避免使用不准确的轮胎模型，反求获得实车试验中的真实激励，提高虚拟仿真模拟精度。并针对汽车零部件和白车身在实车试验中振动冲击特点，分别建立零部件虚拟台架和整车虚拟台架。（2）提出分段加权的频率响应函数的识别方法，提高系统的频率响应函数识别的准确度。系统频率响应函数识别是进行基于真实路谱重现的虚拟试验台架仿真的基础，而汽车在实车试验过程中，存在较多的非线性因素影响，传统识别方法只能获得系统在线性阶段的频率响应函数，导致虚拟台架系统的初始误差较大，导致真实路谱重现难度较大。本文提出分段加权的频率响应函数的识别方法，综合考虑系统的线性和非线性部分影响，来获得系统的新频率响应函数。数值计算案例表明：该方法能够减小系统的初始误差，降低虚拟台架中真实路谱的迭代重现难度。（3）提出自适应迭代的道路模拟控制算法，解决虚拟台架仿真中真实路谱信号重现迭代收敛难题。目前，道路模拟控制算法均采用远程参数控制方法，该方法采用简单修正来获得系统的目标信号迭代重现，导致迭代过程中容易出现发散或收敛效率过低问题。本文提出自适应迭代的道路模拟控制算法，通过自动实时动态调整误差反馈值，解决迭代收敛难题，快速而准确的获得真实路谱的虚拟仿真重现，提高计算效率。（4）针对汽车零部件疲劳寿命预测问题，采用基于真实路谱重现的零部件虚拟台架方法，获得零部件准确疲劳寿命预测结果。零部件承受实车试验中复杂空间运动导致的冲击载荷，直接测试或虚拟仿真的方法均难获得其准确冲击载荷。针对该问题，本文采用基于真实路谱重现的零部件虚拟台架方法，获得在试验中真实冲击载荷，用于疲劳寿命预测，指导其结构设计和疲劳性能提升。零件的疲劳寿命预测与改进案例表明：该方法能够精确虚拟重现零部件在实车试验中振动冲击，从而获得零部件精确的疲劳寿命预测结果，能有效缩短汽车零部件的研发周期，实现汽车零部件的疲劳性能快速提升。（5）针对白车身疲劳寿命预测问题，采用基于真实路谱重现的整车虚拟台架，获得车身结构路准确疲劳寿命预测结果，并针对白车身仿真模型较大而导致的计算效率低问题，采用应力恢复子模型方法运用于白车身结构计算中，减少改进方案计算时间和大幅提高仿真效率。在实车试验中，白车身在实车试验中的复杂冲击载荷，由于轮胎模型、悬架与白车身连接的复杂性，采用传统方法无法获得其精确冲击载荷，导致白车身结构的疲劳寿命预测精度较低。本文采用整车虚拟台架，解决实车试验中白车身冲击载荷求解难题，提高整车疲劳试验虚拟仿真的准确性，获得白车身结构的准确疲劳寿命预测值，指导其结构改进及疲劳性能提升。基于真实路谱重现的整车虚拟台架方法与实车道路试验结果对比表明：采用该方法，能够精确重现整车的路试振动冲击，获得白车身结构的准确疲劳寿命预测，能够有效指导车身结构设计和改进，实现白车身疲劳性能快速提升。更多还原

【Abstract】 As the fatigue of automotive structures determine the life of the car, and thefatigue of vehicle are usually predicted by real tests or computer-aided techniques.The test can provide very accurate results, while the test cycle is very long and thecosts are also enormous; the virtual simulation is subject to major differences betweencomputer-assisted technology and the actual test or fatigue theory, and the predictionaccuracy is low with a limited application in engineering. Currently, the automotivestructural fatigue assessment mainly depends on the actual test, which greatly limit sits research, development efficiency and cost.In this paper, computer aided technology for automotive structural fatigueprediction system is investigated. We propose a virtual test method to improve theaccuracy of virtual simulation, and also apply it to the automotive structural fatiguelife prediction, which also improve the accuracy. As this method can predict theautomotive structural fatigue very quickly and accurately, and guide structural fatigueperformance optimization, and then can reduce the number of the virtual test, researchand development period and cost. This new method has very important theoretical andengineering application value. The main contents and innovations include:(1) Proposed virtual test method based on real road spectrum to improve theaccuracy of virtual simulation of automotive. In the virtual simulation of vehicleusing computer-aided techniques, the relationship of road-tires-vehicle coupling isvery complex, especially for tire model simulation, which limiting t he improving ofaccuracy for fatigue prediction. In this paper, virtual test method based on real roadspectrum is proposed, the iterative feedback is used to correct the virtual simulation.This method avoids the use of inaccurate tire model, which improv es the accuracy ofsimulation. The body-in-white and parts virtual test bench are then establishedaccording to the real characteristics of vibration and shock tests.(2) Proposed sub-weighted frequency response function of identifying method toimprove the system’s frequency response function recognition accuracy. Systemfrequency response function recognition is the basis of virtual test method based onreal road spectrum. In the real test of vehicle, non-linear factors should be considered,while the system frequency response function only considered the linear stage, whichinducing large error of virtual simulation. In this paper, sub-weighted frequency response function of identifying method is proposed. In this method, the newfrequency response function is then obtained by sub-weighted of linear and nonlinearpart. Numerical case shows that the method can decreases the system’s initial error,reduce the difficult of reproducing real road spectrum in virtual bench.（3）Proposed an adaptive iterative algorithm of road simulation to solve thedifficulty of convergence for reproduction virtual simulation using the real roadspectrum. At present, The Remote Parameter Control method is adopted in the virtualroad simulation with a simple iterative correction to obtain the nonlinear system,which inducing the iteration process easy divergence and low efficiency. In this paper,adaptive iterative error feedback control is adopted in the virtual test method based onreal road spectrum, the error feedback will be obtained in each subsequent iteration,which can solve the convergence problems quickly and accurately and reproduce thespectrum of virtual simulation with high computational efficiency.(4) Proposed virtual test method for vehicle parts based on real road spectrum toget the accuracy prediction of fatigue. In the actual driving test, parts are installed inthe car, and exposed to complex spatial impact load, and we cannot get its accuratevalue by direct tests or simulations. To solve this problem, this pa per proposed virtualtest method for vehicle parts based on real road spectrum, and get accurate loadboundary conditions for fatigue prediction, which will guide its structural design anddurability improvement. Numerical method and experiment verified th at it caneffectively shorten the development cycle for automotive parts development andimprove engineering efficiency, realize rapid promotion of durability for automotiveparts.(5) Proposed virtual test method for vehicle body-in-white based on real roadspectrum to get the accuracy prediction of fatigue, and Recovery Based StressSub-model Method using stress recovery is also proposed to reduce the computationsimulation and optimization time, which leads to a substantial increase in efficiency.In the actual vehicle test, the vehicle body structure exposed to the impact load fromthe tire, and suspension, and due to the complexity of the suspension and tire model,the conventional methods cannot get exact impact load for the body structure, thus thefatigue life prediction accuracy is very low. This paper proposed virtual test methodfor vehicle body-in-white based on real road spectrum, which improve vehicle virtualfatigue simulation accuracy and computational efficiency. The accurate bodystructure fatigue prediction can also guide its structural durability optimization.Numerical simulation and experiment verified that it can reproduce the shock and vibration on the road test, realize rapid promotion of durability for vehicle body,which effectively guide the design and optimization of the body structure.更多还原