基于人工智能的机器人折弯机滑块行程研究

Research on the slider stroke of robot bending machine based on artificial intelligence

针对金属板料折弯工艺和更高精度的要求,提出基于人工智能神经网络的机器人折弯新技术。对机器人弯折工艺进行特征参数分析及提取,并建立改进算法的BP神经网络模型;比较了不同机器人折弯训练函数下的性能,建立更加有效的神经网络训练函数。通过经验实测值与改进算法的BP神经网络预测值对比,验证了所确定的机器人折弯的改进BP神经网络能够更加精准确定折弯工艺过程中所需的滑块行程,提高了折弯工艺精度。

To improve the bending process of metal plate and to meet the requirement of higher precision,a new technology of robot bending based on artificial intelligence neural network was proposed. The parameters of robot bending plate process were analysed and extracted,and a back propagation（ BP） neural network model via improved algorithm was established. The performances under different robot bending training functions were compared and a more effective neural network training function was formulated. Comparison was made between the measured result and BP neural network preditting（ forecasting） result. The results indicate that the BP neural network for robot bending can determine the slide stroke required by the bending process more accurately,and can improve the precision of the bending process.