【作者】 高晖；

【导师】 李光耀； 钟志华；

【作者基本信息】 湖南大学 ， 汽车工程， 2008， 博士

【摘要】 计算机仿真技术在汽车设计制造中得到了越来越广泛的应用,特别是在汽车碰撞安全性设计中,仿真技术为缩短开发周期、降低成本起着很重要的作用。计算效率和计算精度是汽车接触碰撞仿真技术的关键。本文对参数反求方法、壳单元和沙漏控制算法、多时间步长算法等影响效率和精度的关键理论和方法进行了系统研究。本文的主要创新点包括以下几个方面:(1)材料参数的精度是影响仿真精度的关键因素。本文利用遗传算法的全局优化的优势,提出了一种基于改进遗传算法IP -μGA(a modified micro genetic algorithm with the strategy of Intergeneration Projection)和计算机仿真技术的参数反求方法,编制了相应的计算程序。为了克服遗传算法计算耗时多的问题,本文提出一种直接从冲压成形过程反求材料参数的方法。试验测量值为冲头的力-位移曲线,而待求的参数包括材料的特性参数和本构参数,采用移动最小二乘响应面法作为优化方法以减少调用正问题计算的次数。为了进一步提高计算效率,本文提出用解析式和有限元仿真结合的方法从冲压过程直接反求材料塑性参数。算例表明本方法在保证精度的前提下大幅度提高了计算效率。(2)壳单元算法是汽车接触碰撞仿真的核心算法之一。针对汽车碰撞和冲压成形过程的特点,本文提出了一种新的壳单元算法,编制了相应程序并集成到自主研发的碰撞仿真软件之中。该算法采用面内一点积分、物理稳定沙漏控制的计算格式,针对物理稳定单元存在的翘曲构型计算不精确和翘曲构型膜锁定问题,提出了两种翘曲构型沙漏控制策略,即:沙漏稳定性因子法、非翘曲影响域法。大量的算例和试验验证表明本文的壳单元算法提高了计算精度和稳定性。(3)对汽车碰撞和冲压成形过程的仿真基本上采用显式时间积分的格式。对于复杂的系统,采用多时间步长积分方法能很大程度地提高计算效率。在深入研究多时间积分方法即子循环法的基础上,本文提出了阻尼子循环法。该方法在常速度子循环法的基础上引入阻尼项以消除由于使用子循环法引起的计算误差和不稳定因素。对刚性连接的子循环方法也进行了深入的研究和处理,以满足碰撞过程仿真的需要。算例表明了阻尼子循环法的有效性。(4)本文提出用遗传算法进行某微车的约束系统改进优化,初始约束系统模型经过试验验证后被用来进行相关参数的优化。对于整车安全性的改进,采用结合结构碰撞安全性分析改进和约束系统遗传算法优化的方法进行。结构改进的好坏最终由车身变形的指标和假人的伤害指标来评判。综上所述,本文在参数反演方法、壳单元和沙漏控制技术、子循环方法和应用仿真技术优化实车安全性方面取得了一些成果,提出了三种材料参数反求的方法,为材料参数的准确获取提供了新的途径;提出了一个新的壳单元计算格式和两种翘曲结构沙漏控制方法,为提高汽车碰撞仿真精度探索了新的方法和途径;提出了阻尼子循环方法,为提高计算效率和减少误差提供了手段;提出了一条应用仿真技术和遗传算法解决微车碰撞安全性问题的切实可行的技术路线。更多还原

【Abstract】 In automotive design, Computer Aided Engineering (CAE) has become a key factor to reduce the development time and the cost of products, especially for the design of vehicle crash safety. In order to improve the performance of the simulation methods for impacting process, studies which focus on the inverse problem, the shell element method, hourglass control ideas, multiple time steps algorithm and crash safety improvement using CAE and optimization methods for a microbus are conducted in this paper.The main innovative points of this paper are given as follows:(1)The precise material parameters play an important role to an effective simulation. An identification scheme to couple the finite element method and IP -μGA(a modified micro genetic algorithm with the strategy of Intergeneration Projection)is developed and programmed. The idea is to match the calculated response with the measured one in least-squares sense. The optimization algorithm chooses the parameters which are putted into the finite element code to find the best coincidence between the observed response and the calculated one. This technique is an alternative approach to the conventional method. An identification scheme for the determination of plastic parameters of blank sheet material is presented. This identification scheme is based on the combination of the finite element method (FEM) and the response surface methodology (RSM). The FEM is employed to calculate the responses of a blank sheet and RSM adjusts the material parameters so that the calculated responses match the measured one in a least-square sense. An example shows that the accurate plastic parameters can be obtained from a sheet forming test by using the method. An identification scheme for characterizing the material plastic property directly by using the results of simulation is presented. A series of equations in which the material parameters become unknown are constructed based on results from the calculation of FEM. The macro genetic algorithm is used to solve these unknown parameters. It is found that the identification scheme presents with high computational efficiency and accuracy.(2)A general four-node shell element with a single point quadrature is developed in this paper. This formulation is based on a physical stabilization approach for the control of spurious zero-energy modes (the hourglass mode). In order to eliminate the over hardening response existing in the warped configurations produced by applying the assumed strain method and the physical stabilization approach, an hourglass stability factor is introduced. In this paper a concept of mini-warped domain is presented and the hourglass forces are calculated based on it. The numerical results show that the developed methods possess high stabilization, high convergence and high accuracy.(3)It can decrease significantly the computation time in transient structural analysis to use subcycling algorithms which permit multiple time steps in explicit integration. Several subcycling algorithms were tested and it was shown that constant velocity subcycling algorithm possesses the best accuracy and stability properties. A new subcycling algorithm based on special treatment of rigid joint and proper choice of damper is proposed. The numerical examples show that the algorithm presents with high computational efficiency.(4)Genetic Algorithm(GA) is employed to solve the problem of constrain system optimization. A Madymo frontal crash model is testified its validity and then is optimized to yield the lowest levels of the occupant injury value and at least satisfy the Chinese legal requirements for frontal crash. It wastes little time to run this problem by using GA, as each simulation time is very little. The accuracy of the method is verified by comparison with optimization results and the test results.更多还原