Welding parameter transfer learning prediction method based on data space conversion

• Main Authors: ZHANG CHENG, DING MENGJIA, XIA HAO, JING YAN, CHEN YUSHAN, LI YAO, TIAN ZHEN, ZHANG YINGFENG, KANG CHENGFEI
• Format: Patent
• Language: Chinese; English
• Published: 08.08.2023
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The technical problems that training data samples of a welding parameter prediction model are insufficient due to the fact that welding process parameters are difficult to obtain, and an existing data expansion method is not suitable for conducting data expansion on welding process data stored in an enterprise database so as to meet the sample requirements of welding parameter prediction are solved. The invention provides a welding parameter transfer learning prediction method based on data space conversion. According to the welding process parameter table data, data space migration is carried out on weldment parameters and welding process parameters in different fields, and parameter adjustment knowledge in the welding process parameters is extracted, so that a parameter set of a small sample target threshold can be effectively expanded; and the prediction precision of the small sample welding process parameter prediction model is improved. 为了解决焊接工艺参数获取困难导致焊接参数预测模型训练数据样本不足，而现有的数据扩充方法却不适用于对企业数据库中保存的焊接工艺数据进行数据