In wire-filling GTAW(gas tungsten arc welding), the significant technical bottleneck in matching the filler wire filling speed with heat source parameters severely restricts further improvements in welding quality and efficiency. To address this technical challenge, this study proposes a method for wire-filling speed-mode recognition in GTAW with adaptive matching to heat source parameters. Through systematic investigation of the coupling behavior between wire-filling speed and arc signals, it is found that as wire-filling speed increases, the position of the wire tip within the arc follows a regular pattern of "edge-center-edge" variation, while the voltage drop period of the arc signal exhibits an inverse relationship with wire-filling speed. Based on the experimental analysis of the weld bead morphology, the wire-filling speed is categorized into four typical modes: insufficient wire filling, intermittent bridging transfer, continuous bridging transfer, and excessive wire filling. Building on this, an empirical mode decomposition-long short-term memory network (EMD-LSTM-NN) model is employed to train and recognize the patterns in the arc signals. The experimental results demonstrate that the proposed method achieves a recognition accuracy of 97.31% for the wire-filling speed modes, providing a reliable theoretical foundation and technical support for adaptive wire-filling speed matching in GTAW.

The robotic welding of 3D complex trajectory fillet welds in medium-thick plates remains a challenging endeavor, particularly in regard to the geometric intricacy of the weld seam and the multi-angle variability of the weld seam trajectory. In the context of such complex weld seams, the capabilities of existing vision sensors may be constrained in terms of high-precision measurement and real-time feedback. In contrast, rotating arc sensing technology offers significant advantages in terms of real-time performance, robustness, and environmental adaptability. To address these issues, we propose a 3D weld tracking method for robotic GMAW welding based on rotating arc sensing. The method mainly proposes a welding torch spatial position recognition method for the welding torch to extract its spatial position information, followed by the application of a pattern recognition algorithm to locate and identify the inflection point of the 3D weld. Specifically, this study introduces a pattern recognition algorithm that utilizes torch tilt change information to develop a suitable weld seam tracking method and corresponding controllers tailored to various tilt characteristics. Finally, the 3D weld seam tracking test results show that the tracking method can accurately identify the position of the welding torch; the average error of the monitored tilt of the welding torch is 3.3°, with a standard deviation of 0.38°, and the average tracking error in the direction of the Y-axis and Z-axis welding trajectory does not exceed 0.27 mm, satisfying the criteria for real-time welding of 3D complex trajectory weld seams. for real-time welding of 3D complex trajectory weld seams.

Wire-feeding height is a crucial parameter in wire-feeding gas tungsten arc welding (GTAW) that significantly affects arc stability and droplet transfer mode, yet effective closed-loop control strategies remain underdeveloped. Therefore, this paper proposes a real-time control method for wire-feeding height in robotic GTAW based on wire sensing and machine learning. By analyzing the influence of wire-feeding height on the voltage signal between the wire and the workpiece under different droplet transfer modes, it was found that during the interrupted bridge transfer mode, the wire-feeding height exhibits a linear correlation with the droplet growth phase in the voltage signal, while during the free flight transfer mode, the wire-feeding height shows a linear correlation with the droplet detachment phase in the voltage signal. Based on these findings, this study innovatively proposes an NBM(Naive Bayes Model)-based feature extraction method, successfully establishing a mathematical model for wire-feeding height deviation adaptable to multiple transfer modes. Furthermore, to accurately identify the droplet transfer mode, a method based on EMD-SVM(Empirical Mode Decomposition and Support Vector Machine) was introduced, achieving a recognition accuracy of 100 %. Finally, an incremental fuzzy PID controller for wire-feeding height was designed and experimentally validated. The results demonstrate that this method can achieve stable control of wire-feeding height, with control errors within 0.2 mm.

3D welds are prevalent in advanced manufacturing sectors such as pressure vessels, nuclear power construction, and aerospace. However, these industries employ colored metals with highly reflective surfaces and utilize Tungsten Inert Gas Shielded Arc Welding (GTAW), necessitating precise welding gun positioning. The complexity of 3D welds significantly burdens offline programming and teaching. Besides, traditional laser vision sensors prove ineffective due to substantial reflections from the workpiece surfaces. At present, no method exists for high-precision, real-time 3D weld path pose extraction. To solve these problems, we propose a real-time extraction method for 3D weld path poses using multi-pole magnetic-controlled GTAW arc sensing. Firstly, a corresponding mathematical model is developed to facilitate this process. Then, a precise method based on Support Vector Regression (SVR) for extracting arc pressure information is proposed to counter the interference from random noise in arc signals during feature point extraction from weld poses. Through a real-time extraction algorithm for 3D weld path postures based on multipole magnetron GTAW arc sensing. Finally, experimental real-time extractions on various typical workpieces demonstrate that the average extraction error is within 1 degrees, with a maximum error not exceeding 2 degrees. The results demonstrate the effectiveness of the proposed method under the intricate conditions associated with the GTAW process.

Plasma arc welding (PAW) is commonly employed for welding medium and thick plates due to its capability of single-side welding and double-side forming. Ensuring welding quality necessitates real-time precise identification of the melting state. However, the intricate interaction between the plasma arc and the molten pool, along with substantial signal noise, poses a significant technical hurdle for achieving accurate real-time melting state identification. This study introduces a magnetically controlled method for identifying plasma arc melt-through, which integrates arc voltage and arc pool pressure. The application of an alternating transverse magnetic field induces regular oscillations in the melt pool by the plasma arc. The frequency characteristics of the arc voltage and pressure signals during these oscillations exhibit distinct mapping relationships with various fusion states. A hybrid feature extraction model combining gray correlation analysis (GRA) and the Pearson correlation coefficient (PCC) is devised to disentangle the nonlinear, non-smooth, and high-dimensional repetitive features of the signals. This model extracts features highly correlated with the fusion state to construct a feature vector. Subsequently, this vector serves as input for the fusion classification model, CNN-SVM, facilitating fusion state identification. The experimental results of melt-through under various welding speeds demonstrate the robustness of the proposed method for identifying melt-through through magnetic field-assisted melt pool oscillation, achieving an accuracy of 96%. This method holds promise for integration into the closed-loop quality control system of plasma arc welding, enabling real-time monitoring and control of melt pool quality.

In the wire-filled GTAW(gas tungsten arc welding), the arc length is prone to random fluctuations due to uneven workpiece surface height and tungsten electrode wear, which disrupts the metal transfer and affects the quality of welding. Realizing real-time control of the arc length is crucial to addressing this issue. However, the current methods for arc length control in wire-filled GTAW do not adapt to the metal transfer mode, making it challenging to overcome the interference caused by metal transfer on arc length control. Therefore, in response to this challenge, this paper proposes a real-time arc length control method for wire-filled GTAW based on EMD-SVM (empirical mode decomposition-support vector machine) adaptive metal transfer mode. An arc length model adaptive to the metal transfer mode is established by analyzing the arc voltage signal characteristics under different metal transfer modes, and a method for extracting arc length feature signals based on NBM (Naive Bayes model) is proposed. To adapt to different metal transfer modes, a real-time recognition method for metal transfer modes based on EMD-SVM is introduced, achieving a recognition accuracy rate of 100

Droplet transfer frequency is a decisive factor in welding quality and efficiency in gas tungsten arc welding (GTAW). However, there still needs to be a monitoring method for droplet transfer frequency with high precision and good real-time performance. Therefore, a real-time monitoring method for droplet transfer frequency in wire-filled GTAW using arc sensing is proposed in this paper. An arc signal acquisition system is developed, and the wavelet filtering method filters out noise from the arc signal. An arc signal segmentation method—based on the OTSU algorithm and a feature extraction method for droplet transition based on density-based spatial clustering of applications with noise (DBSCAN)—is proposed to extract the feature signal of the droplet transition. A new conception of droplet transition uniformity is proposed, and it can be used to monitor the weld bead width uniformity. Numerous experiments for monitoring droplet transfer frequency in real time are conducted with typical welding parameters. This method enables the real-time observation of droplet transfer frequency, and the result shows that the average monitoring error is less than 0.05 Hz.

A vibration-resistant detection method of weld position and gap based on laser vision sensing is proposed in this paper due to the failure problem of automatic weave weld tracking of V-butt welds with gaps due to arc light, molten metal splash, seam gap variations, and inertial vibration of the weave motion in the manufacture of weave gas metal arc welding for pipelines vessels and ships. An improved random sampling consistency algorithm and an adaptive grayscale centroid algorithm are proposed to overcome the interference of arc light and molten metal splash, achieving the simultaneous image detection of weld position and gap. Moreover, a moving polynomial fitting algorithm is proposed to overcome vibration interference in the direction of weave motion and correct the weld position. Finally, the experimental results of Z-weave welding seam tracking of S-curve welds show that the proposed method can significantly reduce the weld tracking error, meeting the practical welding requirements. This study provides a new solution for eliminating the vibration interference of system devices in practical weave welding manufacturing.

To ensure optimal penetration in direct current (DC) argon tungsten arc welding, a method was proposed to recognize the molten pool penetration during molten pool oscillation under alternating cusp-shaped magnetic field (ACSMF). An arc discharge model is established to analyze the oscillation mechanism of the molten pool under ACSMF. The periodic variation of arc shape induces the periodic oscillation to establish a mapping relationship with the fluctuation of arc voltage waveform. At the excitation current of 3 A and frequency of 10 Hz, the arc voltage waveform undergoes abrupt changes at 1.26 V and 1.37 V, corresponding to the critical and the full penetration states, respectively. The errors in identifying moving penetration are 0.8 % and 0.7 %, with a full penetration weld, demonstrating a penetration rate of 100 %, indicating effective penetration recognition. Arc sensing technology for tungsten inert gas (TIG) welding under ACSMF is poised for gradual application in single-side welding and double-side forming of medium and thick plates. It is anticipated to evolve towards real-time identification technology for molten pool penetration monitoring.

The main factors limiting the development of wire-based additive manufacturing (w-AM) are deposition process instability and deposition deviation caused by side feed. In this paper, the device tuning accuracy is introduced into the wire-arc deposition model for the first time, a wire-arc stable deposition model in micro-plasma arc directional energy deposition (mPA-DED) was established, and the process of the wire from contacting the micro-plasma arc (mPA) column to conducting stable metal transfer was carefully analyzed. A sensitivity coefficient describing the link between wire size and device adjustment precision in mPA-DED is proposed. The link between the sensitivity coefficient and the maximum meltable wire feeding speed and the threshold value of the maximum stable deposition speed was quantitatively analyzed, and the model was verified by cross-comparison experiments among the design wire size, arc size, and wire-arc position deviation. A well-formed 40-layer curved metal sample with a material area utilization ratio of 90.54 % was steadily deposited by using the proposed model and the optimized deposition parameters. This study offers theoretical guidance for designing an online control system for the deposition process to improve the stability of the fabrication process and deposition accuracy in micro-plasma arc freeform fabrication.

针对纵向磁场作用下的电弧难提取焊缝信息的问题,设计一种由 3个纵向分布磁感线圈组成的'山'形分布纵向磁场传感器.利用COMSOL软件模拟非对称纵向磁场作用电弧形态.取焊接过程电弧电压分布对应的磁感应强度作为焊缝识别试验的磁感应强度.用高速摄影仪拍摄非对称纵向磁场作用下的电弧运动轨迹,并与新型传感器设计的电弧运动轨迹进行比较,验证纵向磁场传感器产生非对称纵向磁场的电弧形态变化.结果表明,非对称纵向磁场能控制电弧进行焊缝识别,并能解决窄间隙焊接过程中的咬边和侧壁不融合.该方法为磁控焊缝跟踪传感器在窄间隙焊接的应用开辟了新的方向.

Real-time tracking and alignment of the welding torch with the center of the weld seam are critical for automatic root pass welding of medium-thick plates. However, assembly errors, heat input, and interference from droplets on the arc signal cause real-time changes in the weld seam and gap, making real-time tracking of root pass welding with a variable gap highly challenging. In this study, we propose a tracking method of root pass welding with variable gap based on the magnetically controlled arc sensor. First, a coded magnetically controlled arc sensor was developed for the acquisition of weld seam information. Secondly, combined with the simplified model of arc scanning welds, an arc signal compensation method of selectively taking the average based on the Ransac algorithm was proposed. Finally, we proposed an optimized weld deviation detection method by combining the mathematical model of magnetically controlled arc oscillation with the mathematical model of droplet transition size and frequency under the influence of an alternating magnetic field. Weld deviation detection was performed using the compensated arc signal combined with the positioning information provided by the excitation current coding sequence. The experiment results showed that the maximum detection errors in the Y-axis and Z-axis directions do not exceed 0.40 mm and 0.23 mm, respectively. Our proposed method can effectively track the weld seam of root pass welding with a variable gap.

3D折线焊缝大量存在于海工装备、大型起重装备、物流运输装备等制造领域中，属于典型的复杂轨迹焊缝，主要通过示教再现的方式进行自动焊接。大量的重复示教工作严重限制了焊接效率和质量，实现焊缝实时跟踪是提高焊接质量和效率的有效途径。针对3D折线焊缝实时跟踪问题，建立一种基于轨迹在线识别的机器人摆动熔化极气体保护焊（Gasmetalarc welding,GMAW）实时跟踪系统。首先，提出一种基于点云数据处理的3D折线焊缝起焊点焊枪位姿检测方法，获取焊缝起焊点焊枪位姿；利用3D折线焊缝位姿信息在线快速提取方法获取焊接过程中焊缝位姿信息，实现焊缝轨迹在线识别。然后，利用摆动电弧偏差识别方法获取焊缝偏差。最后，提出一种基于轨迹在线识别的3D折线焊缝机器人摆动GMAW实时跟踪方法，利用模糊PID控制方法实现焊缝实时跟踪。针对折角范围为130°～230°的典型3D折线焊缝的焊接试验表明，起焊点寻位位置检测误差小于0.4mm，姿态估计误差小于1.8°，焊缝跟踪误差不超过0.4mm，满足3D折线焊缝实时跟踪的要求。

针对工件上多个不同直径管道的筒体内壁焊缝位置信息难于提取的技术问题,设计了一种能识别支管空间位置和尺寸信息的焊缝位置识别传感器,通过该装置采集到的数据,结合已知工件空间位置和尺寸信息,建立管道插接焊缝的位置模型,并推导出基于此模型的焊缝特征矩阵和焊枪姿态矩阵.将采集的数据结合上述数学模型,在MATLAB软件中进行仿真对比.结果表明,其精度误差最大为0.25 mm,满足实际焊接精度要求,验证了该传感器与数学模型的准确性.该传感器及其焊缝特征识别方法具有通用性,对管道插接焊接任务的自动化、智能化具有重要意义.

Wire-arc additive manufacturing (WAAM) technology realizes part manufacturing with high deposition efficiency and material utilization. However, the unstable metal transfer process, high heat input, and unconcentrated arc energy make this technology face the challenges of poor stability, forming accuracy, and application range. In this work, an innovative micro-plasma arc directed energy deposition (mPA-DED), that is, a local alternating magnetic field (LAMF) to dynamically control the metal transfer to mitigate these issues. Experiments show that LAMF can control the heated position of the liquid bridge, reduce excess energy, and maintain the dynamic stability of the bridge transfer. Metal parts with high forming accuracy can be stable fabricated. In addition, when the wire feeding speed changes, the deposition layer with good appearance can be obtained again by adjusting the frequency of the magnetic field, which indicates the good adapt-ability of this technology. Under the action of LAMF, the dynamic change process of the bridge transfer has a certain periodicity, and the mechanism of this regular change is carefully analyzed for the first time. Finally, a well-formed metal wall has been fabricated by using the novel technology of LAMF to control the mPA-DED.

In the process of high speed gas metal arc welding (GMAW), large box girder has many complex working conditions, such as positioning weld, low assembly accuracy, difficulty in strictly ensuring the pose of workpiece through tooling, real-time change of 3D pose of weld, etc. The vision based weld recognition method has a large amount of calculation and is not aimed at the workpiece with positioning weld, making it difficult to obtain the 3D pose of large box girders quickly. Aimed at this problem, a fast 3D pose estimation method for large box girder based on laser displacement sensing and point cloud clustering was proposed. Using this method, the vertical plane and flat plane of the welding seam of large box girders were obtained by fast segmentation of point cloud. Then, the pose information of welding seam was calculated. A pose information estimation test was conducted for welds with different poses. The results show that when the welding speed is up to 1200 mm/min, the pose error of the weld is less than 0.25 mm and 1.8°respectively. The robustness of the automatic welding of large box girders to complex conditions of positioning weld and low assembly accuracy is enhanced, which greatly improves the welding quality.

Three-dimensional (3D) zigzag-line welding seams are found extensively in the manufacturing of marine engineering equipment, heavy lifting equipment, and logistics transportation equipment. Currently, due to the large amount of calculation and poor real-time performance of 3D welding seam detection algorithms, real-time tracking of 3D zigzag-line welding seams is still a challenge especially in high-speed welding. For the abovementioned problems, we proposed a method for the extraction of the pose information of 3D zigzag-line welding seams based on laser displacement sensing and density-based clustering point cloud segmentation during robotic welding. after thee point cloud data of the 3D zigzag-line welding seams was obtained online by the laser displacement sensor, it was segmented using theρ-Approximate DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm. In the experiment, high-speed welding was performed on typical low-carbon steel 3D zigzag-line welding seams using gas metal arc welding. The results showed that when the welding velocity was 1000 mm/min, the proposed method obtained a welding seam position detection error of less than 0.35 mm, a welding seam attitude estimation error of less than two degrees, and the running time of the main algorithm was within 120 ms. Thus, the online extraction of the pose information of 3D zigzag-line welding seams was achieved and the requirements of welding seam tracking were met.

Traditionally, tidal level is predicted by harmonic analysis (HA). In this paper, three hybrid models that couple varied pre-processing methods, which are empirical mode decomposition (EMD), ensemble empirical mode decomposition (EEMD), and empirical wavelet transform (EWT), with the nonlinear autoregressive networks with exogenous inputs (NARX) were applied to forecast tidal level. The models were, namely, EMD-NARX, EEMD-NARX, and EWT-NARX. The sub-series obtained by using EMD or EEMD or EWT were then used as the input vectors to the NARX with the original data as targets. Notably, the EWT-NARX model was employed to predict the tidal level for the first time. Simulations were based on the measurements from four tidal stations at the Pearl River Estuary, China. The results showed that the EWT-NARX, EEMD-NARX, and EMD-NARX outperformed the HA model. Specifically, EWT-NARX was optimal among the four. Moreover, from the Hilbert energy spectra we can see the EWT solved the mode-mixing problem that EMD and EEMD suffered from, thus enabling precise tidal level prediction. Simulations and experimental results confirmed that the EWT-NARX model can achieve prediction of the tidal level with high accuracy.