基于声发射信号特征的高速焊凝固热裂纹在线检测方法研究

Welding hot crack is one of the most harmful defects in high-speed welding structure. The existing detection methods are unable to conduct the online detection, so the high speed welding technology is diffcult to realize the purpose of high efficiency production of the welding structure. Therefore, this project proposes the online detection method for welding hot crack based on the characteristics of acoustic emission signal. Firstly, the dynamics model of welding hot crack under transient thermal impact is studied to reveal the dynamic fracture behavior and acoustic emission response mechanism of welding solidification hot crack, the mapping relationship between the hot crack fracture parameters and the response characteristics of acoustic emission signal are obtained, the acoustic emission characteristic functions are established as different crack morphology. The acoustic emission wave model is also studied to reveals the time, frequency and time-frequency domain characteristics of acoustic emission signal as well as the propagation. Secondly, the modern signal analysis methods are studied to propose the multi-domain feature extraction methods of acoustic emission signal, and build the sensitive feature vector of welding hot crack with the information fusing of multi-channel acoustic emission signal. Thirdly, based on advanced pattern recognition method, the high precision detection and identification method for the welding hot crack is proposed by the acoustic emission characteristic vector. The project aims to provide the theoretical foundation and technical support for the accurate online detection of welding hot crack, has important significance of academic and application.

凝固热裂纹是高速焊焊接结构危害最大的缺陷之一，现有检测方法无法实现对其快速、准确的在线检测，是制约高速焊接技术实现焊接结构高效化生产的技术瓶颈。项目提出基于声发射信号特征的高速焊凝固热裂纹在线检测方法。首先，研究焊接凝固热裂纹冷却热冲击作用动力学模型，揭示焊接凝固热裂纹冷却过程动态断裂行为及声发射响应机理，探求热裂纹断裂参数与声发射响应特征的映射关系，构建热裂纹热冲击作用下的声发射表征函数；研究焊接凝固热裂纹冷却热冲击作用声发射波动模型，揭示声发射信号时域、频域和时频域特征及其在焊接结构中的传播特性；其次，研究复杂噪声背景下声发射信号微弱特征信息提取方法，构建基于多域、多通道声发射特征信息融合的热裂纹状态敏感特征向量；最后，研究基于声发射特征向量的热裂纹智能识别模型，实现焊接凝固热裂纹状态高精度识别，为实现焊接凝固热裂纹在线、准确的检测提供基础理论和关键技术，具有重要的学术意义和应用价值。

裂纹检测是机械结构焊接生产过程质量保障和安全服役的重要环节。项目开展了基于声发射信号特征的高速焊凝固热裂纹在线检测方法研究，在焊接裂纹声发射产生机理、传播特性、复杂背景噪声多模态声发射信号分离与提取和焊接裂纹声发射源定位与识别方面取得了重要进展。(1)建立了移动点热源作用下含裂纹结构焊接热力模型，计算分析了焊接热力作用下裂纹动态行为，揭示了焊接过程凝固裂纹敏感特性，获得了局部热剧烈作用下裂纹形成及扩展近场参数变化规律，并阐述了声发射产生机理和实现了焊接裂纹声发射源数学表征。(2)建立了移动热源作用下裂纹声发射波动模型，计算分析了焊接移动热源作用下声发射传播特性，为声发射传感器频带选择、安装、测试与分析提供了指导与数据。(3)构建了裂纹声发射信号时频能量特征值，为利用声发射信号进行裂纹长度和深度检测提供了参数指标。(4)研制了横向拘束应力可调实验装置，搭建了焊接过程声发射信号测试平台，实验测试分析了焊接过程各激励源声发射信号成分及频域特性。(5)提出了基于时间序列(AR)-Akaike信息准则(AIC)算法的声发射信号到达时间识别方法，提高了平板结构焊缝裂纹声发射信号到达时间识别精度；提出了基于最小二乘法的单纯形迭代定位算法，实现了平板结构焊缝裂纹声发射源高精度和快速定位；提出基于矩张量反演的裂纹类型识别方法，实现了平板焊缝裂纹声发射源定位与类型识别。(6)将同步压缩小波(SST)和S变换算法应用于焊接裂纹声发射信号分离与特征提取，结合主成分分析(PCA)方法，提出了基于SST和PCA的声发射信号时频特征增强方法，构建了定量表征焊接状态(摩擦、电弧冲击、裂纹)的声发射信号近似熵特征向量。(7)提出了基于隐马尔科夫模型(HMM)-支持向量机(SVM)的焊接裂纹混合智能识别模型，实现了焊接裂纹声发射信号在线检测。发表论文18篇(SCI源刊9篇、EI源刊1篇)，录用SCI源刊1篇，出版专著2部，获授权专利5项、软件著作权2项，培养研究生9名，获湖南省科技进步奖、中国仪器仪表学会科技奖等4项。