船舶吊运机械臂焊接接头超声无损智能检测研究  

ltrasonic non destructive intelligent testing of welding joints in ship crane arm

作　　者：季肖枫 顾娜 陆佳皓 张晖 张崇晖 JI Xiaofeng;GU Na;LU Jiahao;ZHANG Hui;ZHANG Chonghui(College of Intelligent Manufacturing and Information Technology,Jiangsu Shipping Vocational and Technical College,Nantong 226010,China;School of Electrical and Energy Engineering,Nantong University of Scinece and Technology,Nantong 226000,China)

机构地区：[1]江苏航运职业技术学院智能制造与信息学院,江苏南通226010 [2]南通理工学院电气与能源工程学院,江苏南通226000

出　　处：《舰船科学技术》2024年第24期149-154,共6页Ship Science and Technology

基　　金：南通市社会民生科技计划项目(MSZ2023007)。

摘　　要：随着海洋工程领域的扩展与深化,船舶吊运机械臂成为海洋作业的关键技术支持。为了保证机械臂焊接接头的焊接质量,降低海洋作业风险,研究以神经网络为技术支持,设计了点焊接头超声无损检测的缺陷智能识别模型。实验结果表明,研究针对超声无损智能检测模型设计的优化算法,在单峰测试函数上最小值的寻优值为4.804E-11。寻优超体积最大为0.954,与真实前沿解的最小距离为0.203。改进之后的检测模型在F1值上最大可达0.946,损失值最小为0.07,分类检测能力较强。该方法可较好地拟合超声信号特征,有效区分焊接接头的不同缺陷,区分合格焊接与缺陷焊接。研究设计的超声无损缺陷智能检测识别模型有效保证了船舶吊运机械臂的焊接质量,满足海洋工程对船舶吊运机械臂焊接的高要求。With the expansion and deepening of the field of ocean engineering, ship lifting robotic arms have become a key technical support for ocean operations. In order to ensure the welding quality of the robotic arm's welding joints and reduce the risk of marine operations, a defect intelligent recognition model for spot welding joint ultrasonic non-destructive testing was designed using neural networks as technical support. The experimental results show that the optimization algorithm designed for the ultrasonic non-destructive intelligent testing model has a minimum optimization value of 4.804E-11 on the unimodal testing function. The maximum optimized hyper volume is 0.954, and the minimum distance from the true frontier solution is 0.203. The improved detection model has a maximum F1 value of 0.946 and a minimum loss value of 0.07, indicating strong classification detection ability. This method can well fit the characteristics of ultrasonic signals, effectively distinguish different defects in welded joints, and distinguish between qualified welding and defective welding. The research and design of an ultrasonic non-destructive defect intelligent detection and recognition model effectively ensures the welding quality of ship lifting robotic arms, meeting the high requirements of marine engineering for the welding of ship lifting robotic arms.

关 键 词：反向传播神经网络 果蝇优化算法 船舶吊运机械臂 超声检测 

分 类 号：TP241[自动化与计算机技术—检测技术与自动化装置]