管材数控绕弯回弹实验研究及BP网络预测模型

The experimental study on the springback and BP NN prediction model of NC rotary-bending of tube

回弹是影响管材数控绕弯成形精度的主要因素之一。在EATON VB100弯管机上对直径φ6mm～φ15mm之间的3种管材进行弯曲成形实验后,利用激光测量仪测量回弹角度,整理得到准确、可靠的回弹实验数据作为样本数据。采用最小二乘法对实验数据进行拟合,得到关于弯曲角度和回弹角度的函数方程,同时分析了相对弯曲半径、相对管径和管材性能对管材弯曲回弹的影响规律,根据实验,确定了影响管材弯曲回弹主要工艺参数。综合主要工艺参数对管材弯曲回弹的交叉影响,以实验数据作为样本数据,建立和优化了BP神经网络模型,预测弯曲角度和相对管径对回弹的影响规律,通过实验数据的对比,证明预测BP模型具有较高的精度。

Springback is one of the main factors effecting on the precision of tube at NC rotary-bending. Tubes of three different kinds of material with diameter varied from 6mm to 15mm were tested on EATON VB100 bending machine. Exact and credible angular springback data were obtained by using laser measuring machine. The functional equation relating the ber,ding angle and the springback was obtained based on the method of least squares. The effects of the relative bending radius, the relative diame ter and tbe material properties on springback were analyzed. The main processing parameters effecting on springback were ascer- tained. Taking experimental data as sample data, BP NN model was established and optimized to analyze the reciprocal effect of main processing parameters on the springback of tube. Comparison between the analyzing results and the experimental data show that the accuracy of the BP model is quite good.