基于BP神经网络的板料折弯件下料尺寸的计算方法

在板料成形工艺中 ,弯曲成形是一种主要的工艺方法。而板料下料尺寸将直接影响到零件的成形精度以及后续工序的成败。针对工厂积累的大量生产和实验数据 ,提出一种基于人工神经网络计算弯曲件板料展开长度的方法。在概述解析法确定板料展开长度的基础上 ,建立了计算展开长度的人工神经网络模型。以板料厚度、成形半径及成形角度作为输入参数 ,中性层位移系数作为输出参数 ,用已积累的大量数据作为训练样本 ,对神经网络进行训练 ,得到这些参数和中性层位移系数的隐性规律 ,实现了弯曲成形板料展开长度的快速计算。实验结果表明 ,利用人工神经网络能够快速准确地计算出板料展开长度。

Research of the bending part blank calculation based on a back-propagation neural network

In processes of sheet metal forming, the bending is a main process. The developed length is very important to precision of the sheet metal products and success of subsequent handling. A new method based on artificial neural network to calculate the developed length quickly is proposed according to many data accumulated during experimenting and manufacturing. Artificial neural network model of calculating the developed length is established on basis of its analytic method. The network is trained by sample data composed of the sheet metal thickness, bending radius, and bending angle which constitute input parameters and displacement coefficient of neutral fiber which is output parameter. Latent regularity of these parameters and displacement coefficient of neutral is gained. So the developed length is calculated quickly and precisely. Experimental result shows the developed length of sheet metal can be calculated quickly and précised using artificial neural network.