王凯,卢楚文,易江龙,等.基于BAS算法优化的电弧增材制造焊道尺寸预测[J].精密成形工程,2024,16(4):190-199.
WANG Kai,LU Chuwen,YI Jianglong,et al.Bead Size Prediction for Arc Additive Manufacturing Based on BAS Algorithm Optimization[J].Journal of Netshape Forming Engineering,2024,16(4):190-199. |
| 基于BAS算法优化的电弧增材制造焊道尺寸预测 |
| Bead Size Prediction for Arc Additive Manufacturing Based on BAS Algorithm Optimization |
| 投稿时间：2023-11-14 |
| DOI：10.3969/j.issn.1674-6457.2024.04.023 |
| 中文关键词: 冷金属过渡弧焊(CMT) 焊道尺寸 天牛须算法 BP神经网络 预测模型 |
| 英文关键词: cold metal transfer welding (CMT) weld bead size beetle whisker algorithm back propagation neural network (BPNN) prediction model |
| 基金项目:广东省自然厅省级促进经济高质量发展(海洋经济发展)海洋六大产业专项项目(粤自然资合[2023]32号);阳江市人才项目(RCZX2022018);广东省基础与应用基础研究基金项目(2021A1515011756);广东省基础与应用基础研究基金项目(2022A1515010761);广东省高校现代陶瓷与铝型材装备重点实验室(2017KSYS012) |
|
| 摘要点击次数: 885 |
| 全文下载次数: 482 |
| 中文摘要: |
| 目的 为提高实际应用中电弧增材制造对工艺参数的选取效率及成形形貌的控制效果，建立高效且精准的成形尺寸预测模型，实现对焊道尺寸的合理预测。方法 在单层单道CMT电弧增材制造实验的基础上，建立基于天牛须搜索算法(Beetle Antennae Search，BAS)优化BP神经网络的焊道尺寸预测模型，利用BAS算法实现对BP神经网络初始权值和阈值的优化，可以实现预测不同工艺参数(焊接速度、送丝速度、干伸长)下焊道的成形尺寸(熔宽、余高)。利用试验验证BAS-BP预测模型的性能，与现有模型进行对比，结果 结果表明该模型具有较高精度的预测效果，能够有效映射工艺参数与焊道尺寸之间的非线性关系，印证了该模型具有良好的拟合和泛化能力，同时其对焊道熔宽和余高的预测误差分别不超过0.2、0.12 mm，预测平均误差率均不超过6%，相对于其他预测模型表现出较好的准确性和稳定性。结论 BAS-BP神经网络预测模型的输出误差较小，网络训练收敛速度加快，避免了过拟合及欠拟合的风险，有效提高了预测模型的泛化能力和预测精度，可以实现一定工艺参数范围内的焊道尺寸预测，为后续电弧增材的实时预测及控制参数应用提供了技术支持。 |
| 英文摘要: |
| The work aims to establish an efficient and accurate forming size prediction model to achieve reasonable prediction of weld bead size so as to improve the selection efficiency of process parameters and shape control effect of arc additive manufacturing in practical application. On the basis of the experiment of single-channel CMT arc additive manufacturing, a solder path size prediction model optimized by the BP neural network based on the Beetle Antennae Search (BAS) algorithm was established, and the initial weights and thresholds of BP neural network were optimized by the BAS algorithm. It could predict different process parameters (welding speed, wire feed speed, dry extension) corresponding to the forming size of the lower pass (melting width, residual height). The performance of the BAS-BP prediction model was verified by experiments, and compared with the existing model. The results showed that the model had a relatively high precision prediction effect, and could effectively map the nonlinear relationship between process parameters and weld pass size, which confirmed that the model had good fitting and generalization ability. At the same time, the prediction errors of weld width and residual height were less than 0.2 mm and 0.12 mm respectively, and the average error rate of prediction was less than 6%, which showed better accuracy and stability than other prediction models. The BAS-BP neural network prediction model has small output error and accelerated convergence speed. It avoids risks of overfitting and underfitting, and effectively improves the generalization ability and prediction accuracy of the prediction model, and can realize the weld size prediction within a certain range of process parameters, providing technical support for the real-time prediction and control parameter application of the subsequent arc additive. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
渝公网安备 50010702501719号