基于RBF神经网络的变压边力优化研究

发布时间：2018-11-21 08:04

【摘要】：冲压成形是一项十分重要的零件制造方法,在汽车、飞机等工业领域得到了广泛的应用,是工业制造领域中的重要组成部分。对于一些形体表面极为复杂的零部件来说,在生产过程中需要通过调整压边力大小来严格控制板料不同成形阶段或区域位置的流动,从而减少甚至消除因流动速度差别过大而造成的拉裂、起皱和回弹等成形缺陷。在一个新产品的开发过程中,需要通过反复地试模来获得满足生产要求的模具,其中经常会因人为因素的缘故而造成整个模具的报废,这无形中就增加了制模成本和周期。将数值模拟和近似模型优化技术应用到板料成形中,不仅可以极大缩短新产品的开发周期,而且能够准确获得最优工艺参数组合。基于此,本文结合数值模拟和近似模型优化技术,对板料成形中压边力进行了如下研究：首先,在充分考虑板料成形拉裂、起皱和回弹等多目标情况下,采用灰色关联分析的方法对影响板料成形的工艺参数进行关联分析,获得了各个因子与板料成形质量的关联度,通过比较关联度,验证了压边力的控制对提高板料成形质量的重要性。其次,基于人工免疫算法,在保证种群多样性的同时,为了提高收敛速度,将适应度概率与浓度抑制概率相结合,并加入精英交叉,改进了人工免疫算法搜索性能。依据RBF神经网络的基本原理,将人工免疫算法用于RBF神经网络训练中,获得了较优的中心和宽度参数,建立了一种基于人工免疫算法的RBF神经网络近似模型。最后,以NUMISHEET标准考题中的方盒和S梁作为研究对象,以变压边力作为设计变量,以板料成形后最大增厚、最大减薄为成形质量指标,利用拉丁超立方进行抽样,使用仿真软件Dynaform进行成形仿真获得训练样本。以人工免疫算法为RBF神经网络的训练方法,分别建立随行程、时间和位置变化的变压边力与成形质量之间的近似模型,采用人工免疫算法对该模型进行优化,获得最优变压边力。通过与恒定压边成形进行比较,表明板料成形中采用变压边力能够有效提高成形件的成形质量。
[Abstract]:Stamping forming is a very important part manufacturing method, which has been widely used in automobile, aircraft and other industrial fields. It is an important part in the industrial manufacturing field. For some parts with extremely complicated shape surface, the flow of sheet metal at different forming stages or regions should be strictly controlled by adjusting the size of the blank holder force during the production process. Thus reducing or even eliminating the forming defects such as crack wrinkle and springback caused by excessive flow velocity difference. In the process of developing a new product, it is necessary to repeatedly try the mold to obtain the mould that meets the production requirements, which often results in the die scrapping due to human factors, which increases the cost and cycle of the mould making intangibly. The application of numerical simulation and approximate model optimization techniques to sheet metal forming can not only greatly shorten the development period of new products, but also accurately obtain the optimal process parameters combination. Based on this, this paper combines numerical simulation and approximate model optimization technology to study the blank holder force in sheet metal forming as follows: firstly, considering the multi-objective of sheet metal forming such as crack, wrinkle and springback, etc. By using the method of grey correlation analysis, the correlation analysis of process parameters affecting sheet metal forming is carried out, and the correlation degree between each factor and forming quality of sheet metal is obtained, and the correlation degree is compared by comparing the correlation degree between each factor and the forming quality of sheet metal. The importance of the control of blank holder force to the improvement of sheet metal forming quality is verified. Secondly, based on the artificial immune algorithm, in order to improve the convergence speed and ensure the diversity of the population, the fitness probability is combined with the concentration suppression probability, and the elite crossover is added to improve the search performance of the artificial immune algorithm. According to the basic principle of RBF neural network, the artificial immune algorithm is applied to the training of RBF neural network. The optimal center and width parameters are obtained, and an approximate model of RBF neural network based on artificial immune algorithm is established. Finally, taking the square box and S beam in NUMISHEET standard test as the research object, taking the variable blank holder force as the design variable, taking the maximum thickening and thinning of the sheet metal after forming as the forming quality index, the Latin hypercube is used for sampling. Simulation software Dynaform is used to obtain training samples for shaping simulation. Using artificial immune algorithm as the training method of RBF neural network, the approximate models between variable blank holding force and forming quality with travel, time and position are established, and the model is optimized by artificial immune algorithm. The optimal variable blank holder force is obtained. By comparing with the constant blank holder forming, it is shown that the variable blank holder force can effectively improve the forming quality of the forming parts in sheet metal forming.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TG386

【参考文献】