Xu Jinhan, Yan Jun, Huang Yan, Ding Dawei
College of Automation & College of Artificial Intelligence, Nanjing University of Posts and Telecommunications, Nanjing 210023, China.
Jiangsu Yawei Machine Tool Co., Ltd., Yangzhou 225200, China.
Sensors (Basel). 2024 Dec 9;24(23):7863. doi: 10.3390/s24237863.
Amidst the accelerating pace of automation in sheet metal bending, the need for small-batch, multi-varietal, efficient, and adaptable production modalities has become increasingly pronounced. To address this need and to enhance the efficacy of the bending process, this study presents the design and development of an embedded soft PLC (Programmable Logic Controller) rooted in the Codesys development platform and leveraging the ARM Cortex-A55 architecture. This controller employs the EtherCAT communication protocol to facilitate seamless and efficient interactions with fully electric servo-driven CNC (Computerized Numerical Control) bending machinery. To mitigate the challenge of bending springback errors, a finite element simulation model is constructed and refined through the application of ALE (Arbitrary Lagrangian-Eulerian) adaptive grid technology, thereby bolstering simulation precision. Subsequently, an enhanced WOA-BP (Whale Optimization Algorithm-Backpropagation) model, integrating Latin hypercube sampling and neural network techniques, is deployed to anticipate and counteract these springback errors. Experimental outcomes demonstrate that the proposed methodology effectively constrains the final forming angle deviation to within 0.3°, significantly enhancing the reliability and precision of the bending system. This achievement not only underscores the technical feasibility but also contributes to advancing the frontier of sheet metal bending automation.
在钣金弯曲自动化进程加速的背景下，对小批量、多品种、高效且适应性强的生产模式的需求愈发凸显。为满足这一需求并提高弯曲工艺的效率，本研究介绍了一种基于Codesys开发平台并采用ARM Cortex - A55架构的嵌入式软可编程逻辑控制器（PLC）的设计与开发。该控制器采用EtherCAT通信协议，以便与全电动伺服驱动的计算机数控（CNC）弯曲机械实现无缝且高效的交互。为应对弯曲回弹误差的挑战，构建了有限元仿真模型，并通过应用任意拉格朗日 - 欧拉（ALE）自适应网格技术对其进行优化，从而提高仿真精度。随后，部署了一种集成拉丁超立方采样和神经网络技术的改进型鲸鱼优化算法 - 反向传播（WOA - BP）模型，以预测并抵消这些回弹误差。实验结果表明，所提出的方法有效地将最终成型角度偏差控制在0.3°以内，显著提高了弯曲系统的可靠性和精度。这一成果不仅强调了技术可行性，也有助于推动钣金弯曲自动化的前沿发展。
