返回
类型 综合研究 预答辩日期 2017-12-12
开始(开题)日期 2014-07-01 论文结束日期 2017-10-17
地点 东南大学九龙湖校区机械楼327 论文选题来源 中央、国家各部门项目     论文字数 7.4 (万字)
题目 薄板的折弯回弹及拉深成形预测模型的研究
主题词 折弯回弹,拉深成形,预测模型
摘要 折弯和拉深是钣金成形领域中应用较为广泛的加工方法,非常适合进行大批量零件的生产。为了避免在进行大批量零件生产时出现大范围的零件加工缺陷,需要对折弯和拉深成形工艺进行深入的研究。本文以折弯和拉深成形为研究对象,对折弯板材参数的反求、折弯回弹的预测、杯形件二次拉深的应力分析及杯形件成形质量的预测等进行了研究,并开发了适用于数控设备集成应用的钣金成形预测系统。本文为数控折弯和成形设备提供了必要的算法和技术上的支持,对于相关产品的开发具有重要意义。 联合径向基函数法和粒子群优化算法,开发了一种动态近似优化模型。其相对于静态近似优化方法具有计算量小和计算精度高的优点。将其作为优化算法,以折弯回弹的实验数据为基准,联合折弯回弹的有限元模拟结果,以实验和模拟结果之间偏差最小为优化目标,对B340LA钢板的塑性参数进行了反求,反求的参数通过与真实的材料参数进行对比,显示出较高的反求精度。本文通过折弯实验数据对材料参数进行反求的方法,为获取材料的塑性参数提供了一种高效而简单的实现方式。 对折弯回弹补偿算法进行了研究,目的是为了获取类似于DELEM系统的折弯角预测模型。通过一系列的公式推导和整理,获得了较为实用的回弹折弯角计算模型,由于在推导的过程中假设较多,而且引入了经验公式项,使得计算模型与DELEM系统之间偏差较大。因此,以DELEM系统的测试数据为基准,对计算模型经验公式中的相关参数进行了反求,根据反求参数的曲线表现形式,给出了与其相对应的参数计算公式,并将其带入原计算公式进行替换,如此即对原计算模型进行了改进。改进后的计算模型基本达到了DELEM系统的精度水平。 考虑到误差反馈神经网络(BPNN)具有较强的数据拟合能力,联合平滑的样条曲线函数(Spline),开发了折弯回弹BPNN-Spline预测模型。以精确的有限元模拟结果为基准,应用正交试验方法,建立了BPNN的训练样本集,最终获得了精确的BPNN-Spline预测模型。其相对于传统的BPNN模型具有两大优势:较易获得精确的BPNN训练结果和折弯角与上模下行位移之间单调递减的关系比较容易保证。通过对长度比例缩放系数的影响分析,得到了折弯角与长度比例缩放系数无关的结论,由此将BPNN-Spline 模型的应用范围进行了扩展,BPNN-Spline模型具有了较强的通用性。BPNN-Spline模型对于各种工况下的折弯都表现出较高的预测精度。 考虑到二次拉深成形理论方面成果严重匮乏的研究现状,本文对杯形件二次拉深时的应力变化情况进行了分析和讨论,获得了面内径向应力随凸模拉深行程变化的分析计算模型。以此为基础,获得了凸模拉力的计算公式。分别通过有限元模拟和实验的方法,对其计算精度进行了检验,分析模型显示出较高的计算精度,仅在某些特殊的位置出现了偏差较大的情况。这是由于分析模型考虑的较为理想化,与二次拉深成形不完全一致所致。通过成形参数的影响分析可以发现,减小板料的初始半径,增大杯形件直壁半径、压边圈圆角半径和凹模圆角半径都能够减小最大凸模作用力。 为了在不同的材料和尺寸条件下,快速准确的预测杯形件的成形质量,联合BPNN和Spline开发了用于杯形件成形质量预测的BPNN-Spline模型。其具有较强的通用性、较高的预测精度和较快的计算速度,非常适合作为一种控制算法应用于数控设备当中。通过与各种拉深工况下的有限元模拟和实验测试实例的对比,预测模型都显示出较高的预测精度。以杯形件成形质量BPNN-Spline预测模型作为求解器植入遗传算法,开发了用于杯形件成形参数优化求解的优化模型,并通过优化实例对其进行了测试,优化模型显示出较高的优化效率和优化精度。 为了方便用户应用本文的折弯和拉深成形预测模型,开发了钣金成形的预测系统。当前版本的系统能够实现对折弯角的预测、杯形件成形质量的预测和杯形件成形参数的优化等功能。系统通过调用MATLAB函数程序实现预测计算功能,应用界面和计算内核相互分离,有利于系统后期的维护。钣金成形预测系统对于提升相关企业的生产效率和技术水平都有积极意义。
英文题目 RESEARCH OF PREDICTION MODEL ON SHEET BENDING SPRINGBACK AND DEEP DRAWING
英文主题词 bendingspringback,deepdrawing,predictionmodel
英文摘要 Bending and drawing are widely used in sheet metal forming field. They are very suitable for the mass production. In order to avoid large-scale part defects in the productive process, it is necessary to make a deeply study on bending and drawing process. In this paper, bending and drawing processes were treated as research object, then, the inverse determination of bending material parameter, the prediction of bending springback, the stress analysis on secondary deep drawing process of cup-shaped part and the prediction of cup shaped part forming quality were studied, and a sheet metal forming prediction system suitable for the integration of numerical control equipment was developed. This paper provides the necessary algorithm and technical supports for numerical control bending and forming equipments, which has great significances to develop related products. A dynamic approximation optimization approach was proposed by the radial basis function and particle swarm optimization. In comparison with static approximation optimization method, the dynamic approximation optimization approach has the advantages of less labor consumption and higher calculated accuracy. Using the dynamic approximation optimization approach, based on the combination of bending experimental and finite element simulated results, the plastic parameters of B340LA steel plate were inverse determined. The inverse determined material parameters show better accuracy in comparison with the actual one. It is an efficient and simple way to obtain the material plastic parameters as the inverse determination approach of this paper. In order to obtain the bending angle prediction model similar to the DELEM system, the bending springback compensated algorithm was researched, and a practical bending springback angle calculation formula was deduced out. Due to the existence of assumption and empirical formula, the deviation between calculation formula and DELEM system was larger. Some parameters in the empirical formula of calculation formula were inverse determined based on the test data of DELEM system. According to the curve expression of the inverse determined parameters, the corresponding empirical formula was obtained. Then, the calculation formula was improved when replace the empirical formula with the new one. The improved calculation formula has achieved the precision level of DELEM system basically. Considering the strong data fitting ability of error feedback neural network (BPNN), a springback bending angle prediction model on the combination of error back propagation neural network and spline function (BPNN-Spline) is presented. An orthogonal experimental sample set for training BPNN-Spline is obtained by precision finite element simulation, and the BPNN-Spline black box function of bending angle prediction is established. The BPNN-Spline has advantages on two aspects: accurate BPNN training results and monotonically decreasing relation between bending angle and punch displacement are easier to guarantee. The application range of BPNN-Spline model was extended, because of the independent relationship between bending angle and length zooming factor, then, the versatility of the model was enhanced. The BPNN-Spline model shows higher prediction accuracy under various bending working conditions. Because the theoretical research of secondary deep drawing process is serious shortage, a stress analysis model for the secondary deep drawing process of a cup-shaped part was represented in this paper. The radial stress analysis model taking the punch displacement as the independent variable was obtained. Then, the punch force calculation method was obtained. Through the comparison with finite element simulated and experimental results, the calculation model shows better calculated accuracy, only a little deviation was occurred at some special locations. The deviation was caused by the more idealized analysis model which is incomplete agreement with the actual secondary deep drawing process. Through the impact analysis of forming parameters, it can be observed that the maximum punch force will be decreased by the decreasing initial blank radius, increasing cup-shaped straight wall radius, round corner radius of blank holder and round corner radius of die. In order to rapidly and accurately predict forming quality of cup-shaped part under diverse material and working condition, a BPNN-Spline forming quality prediction model was proposed in this paper. This prediction model is very suitable to be applied in the numerical control equipment as a control algorithm, due to its better versatility, higher prediction precision and faster calculation speed. The model shows better prediction ability in comparison with the finite element simulated and experimental test examples. Taking the prediction model as the solver, an optimization approach of cup-shaped part process parameters was proposed. Through the test of optimization example, the optimization approach shows higher optimization efficiency and precision. In order to the users conveniently use the prediction models of bending and forming, a sheet metal forming prediction system was developed. The current version of prediction system can predict the bending angle, predict the cup-shaped part forming quality, optimize the cup-shaped part process parameters, and so on. The sheet metal forming prediction system can implement the prediction calculation by calling the MATLAB function program. It is beneficial to the later maintenance work of system due to the independent relationship between the application interface and calculation kernel, and it has great significances to promote the efficiency and technical level of related enterprises.
学术讨论
主办单位时间地点报告人报告主题
汤文成教授研究生课题组 2014/11/20 机械楼308 包达飞 高速滚珠丝杠实验台参数识别
汤文成教授研究生课题组 2014/10/30 机械楼308 郭哲锋 动态RBF代理模型对临界压边力的数值逼近
汤文成教授研究生课题组 2016/05/12 机械楼308 郭哲锋 杯形件二次拉深的应力分析
汤文成教授研究生课题组 2015/10/15 机械楼308 包达飞 高速滚珠丝杠进给系统积分滑模控制研究
汤文成教授研究生课题组 2016/11/29 机械楼308 石勇 高速滚珠丝杠进给系统的轨迹跟踪控制方法研究
汤文成教授研究生课题组 2017/01/06 机械楼308 王登铭 轴向柱塞泵故障诊断和预测方法研究
汤文成教授研究生课题组 2013/03/14 机械楼308 郭哲锋 板料折弯回弹的自动补偿技术研究
汤文成教授研究生课题组 2015/09/23 机械楼308 郭哲锋 基于动态优化方法的折弯回弹B340LA板材参数反求
     
学术会议
会议名称时间地点本人报告本人报告题目
2017 8th International Conference on Mechatronics and Manufacturing (ICMM 2017) 2017/01/20-2017/01/22 日本 东京 The Limiting Sheet Diameter Prediction Model for Cup-Shaped Part Drawing Process with Diverse Mould Assemblage Based on RSM
2017年机械工程全国博士生学术论坛 2017/05/10-2017/05/12 安徽 芜湖 Bending angle prediction model based on BPNN-Spline in air bending springback process
     
代表作
论文名称
Bending Angle Prediction Model Based on BPNN-Spline in Air Bending Springback Process
The limiting sheet diameter prediction model for cup-shaped part drawing process with diverse mould
Stress analysis of cup-shaped parts in secondary deep drawing
 
答辩委员会组成信息
姓名职称导师类别工作单位是否主席备注
张琦 正高 教授 博导 解放军理工大学
李东波 正高 教授 博导 南京理工大学
陆宝春 正高 教授 博导 南京理工大学
陈南 正高 教授 博导 东南大学
幸研 正高 教授 博导 东南大学
      
答辩秘书信息
姓名职称工作单位备注
陈斌 其他 讲师 东南大学