Weld defect detection and control is of great significance for the gas tungsten arc welding process of aluminum alloy, and the recognition of penetration states based on acoustic signal has been a hot topic. However, previous studies have ignored that acoustic features have distinct effects on the penetration state. In this article, inspired by the attention mechanism in the vision, a novel auditory attention model combining the attention mechanism with long short-term memory (LSTM) network for penetration state recognition has been proposed. First, the 15-dimensional (15-D) informative features, including 9-D time-domain and 6-D wavelet features, related to the penetration state are extracted. Second, the auditory attention mechanism is explored and the attention-based LSTMs, including attention mechanism before LSTM (AT-LSTM) and attention after LSTM (LSTM-AT), are established, which are experimentally verified to show greater performance than the other traditional methods with an average accuracy of 93.56% and 95.32%, respectively. Furthermore, the attention vectors are visualized to figure out the mechanism of auditory attention when different penetration occurs.

Due to the high positioning accuracy and relatively low prices, SCARA robots are widely used in industrial fields. The objective of this paper is to propose a hand-eye self-calibration algorithm for SCARA robots which could consider both accuracy and computational cost. The previous global optimal hand-eye calibration algorithms based on branch-and-bound (BnB) optimization is limited by their expensive computational cost. The speed of these algorithms depends on the volume of the search space to a large extent, which is the main concern in this paper. Instead of searching over the 3-dimensional parameter space corresponding to the rotation component of hand-eye pose, a new 2-dimensional search space is defined by separating and coupling some calibration parameters by means of the special structure of SCARA robots, which have 4 degrees of freedom (DoFs) including three translation DoFs and only one rotation DoF. The simulation and real experiments show the similar accuracy but much faster speed of the proposed algorithm compared with previous optimal algorithms based on BnB.

Monitoring and controlling of weld joint penetration are essential issues in variable polarity plasma arc welding (VPPAW). In this paper, we develop a flexible visual sensor system to measure the backside keyhole characteristic parameters such as keyhole length, width, and area. Further, data analysis from dynamic welding experiments reveals a nonlinear correlation of the keyhole features and the backside beam width. To provide accurate feedback information, an extreme learning machine method is applied to predict the weld width with sufficient accuracy and less computing time. Then a novel model-free adaptive control (MFAC) is designed to control the weld penetration evaluated by the weld width. Closed-loop experimental results confirm that the MFAC system can simultaneously adjust the welding current and plasma gas flow rate to control the VPPAW process for obtaining a full-penetrated weld under various initial welding conditions and disturbances.

Aluminum alloy is the main structure material in aerospace industry. Online defect detection for aluminum alloy in pulsed gas tungsten arc welding (GTAW) is still challenging, especially for increasing application of robotics. This paper presents an intelligent methodology for real-time evaluation of weld penetration defects based on arc audible sound sensing for aluminum alloy in robotic-pulsed GTAW. The generation mechanism of arc sound was investigated using correlation analysis, high-speed camera observing and frequency spectrum analysis before denoising of arc sound. Two feature selection approaches based on Fisher distance and principal component analysis (PCA) were developed to select the frequency components related to seam defects, and then, their performance were qualitatively and quantitatively analyzed. Finally, a new classification model integrating support vector machine with grid search optimization and cross-validation (SVM-GSCV) was established to identify underpenetration, normal penetration, and burning through. The proposed methodologies were verified to be effective with high accuracy and robustness. This paper can provide some guidance for condition monitoring of additive manufacturing (AM) or process industry.

This paper analyzes the motion planning of a custom-built all position autonomous mobile welding robot. It proposes a novel motion planning method that allows the robot to complete fillet seam welding while the vehicle turns around the right angle seam simultaneously. The global optimal solution is obtained through model modification.

Shape from shading (SFS) is an important domain in computer vision. At first the paper introduces a method of shape from shading based on single image and puts forward its implementation. Then adds proper boundary conditions and adjusts the factor of brightness error function adaptively according to the results by the original algorithm, and finds accuracy of the improved algorithm by the synthesized image. At last the improved algorithm is applied to the surface height reconstruction of weld successfully, and the computing is accurate enough for the inspection on shape of weld.

3D vision technology is one important branch of machine vision, whose aim is to derive an accurate description and measurement of object in 3D scene from 2D image. At present, it is an important research domain in robot, measurement and artificial intelligent fields. The automatization, roboticization and intelligentization are the development trends of welding technology. The vision sensing method is the key to realize the automatic welding. The research of machine vision has been applied in welding field and made great progress. The 3D vision technology and its application research in welding are introduced in this paper. Finally the further development of 3D vision in welding is prospected.

In this paper, a new model is designed for welding flexible manufacturing system (WFMS) of multiple robots in a different way with Petri net (PN), which is multi-robot WFMS time Petri net (MRWTPN). Because the complexity of information and the device of welding system, the function of inhibit arc is imported in the model and we design two-layer way in model's structure, one is main net that depicts the system, the other is subnet depicting appointed device which are welding robot and assistant robot. Hence the control model's characters are proved correctly in theory, too. The model and control's policies researched have established foundation for further intelligent real-time control on WFMS, and it extends the applied area of PN.

We address the problem of control of uncertain systems with time delays. Using the fuzzy logic control and artificial neural network methodologies, we present a self-learning fuzzy neural control scheme for general uncertain processes. In this scheme, a neural network compensator is designed instead of the classical Smith predictor for attenuating the adverse effects of time delays of the uncertain systems. The scheme has been used in control of welding pool dynamics of the arc welding process, and the experiment results show the control scheme available.