Keyhole tungsten inert gas (keyhole TIG) welding is renowned for its advanced efficiency, necessitating a real-time defect detection method that integrates deep learning and enhanced vision techniques. This study employs a multi-layer deep neural network trained on an extensive welding image dataset. Neural networks can capture complex nonlinear relationships through multi-layer transformations without manual feature selection. Conversely, the nonlinear modeling ability of support vector machines (SVM) is limited by manually selected kernel functions and parameters, resulting in poor performance for recognizing burn-through and good welds images. SVMs handle only lower-level features such as porosity and excel only in detecting simple edges and shapes. However, neural networks excel in processing deep feature maps of “molten pools” and can encode deep defects that are often confused in keyhole TIG. Applying a four-class classification task to weld pool images, the neural network adeptly distinguishes various weld states, including good welds, burn-through, partial penetration, and undercut. Experimental results demonstrate high accuracy and real-time performance. A comprehensive dataset, prepared through meticulous preprocessing and augmentation, ensures reliable results. This method provides an effective solution for quality control and defect prevention in keyhole TIG welding process.

Heat input, a crucial factor in the optimization of high-temperature thermocouple laser welding, has a significant impact on the appearance and mechanical properties of dissimilar welded joints involving stainless-steel- and nickel-based alloys. This study focuses on laser overlay welding of austenitic stainless steels and nickel-based alloys. The findings indicate that an increase in heat input has a more pronounced effect on the penetration depth and dilution rate. Under high heat input, the weld has cracks, spatter, and other defects. Additionally, considerable amounts of chromium (Cr) and nickel (Ni) elements are observed outside the grain near the crack, and their presence increases with higher heat input levels. Phase analysis reveals the presence of numerous Cr₂Fe₁₄C and Fe₃Ni₂ phases within the weld. The heat input increases to the range of 30–35 J/mm, and the weld changes from shear fracture to tensile fracture. In the center of the molten pool, the Vickers hardness is greater than that of the base metal, while in the fusion zone, the Vickers hardness is lower than that of the base metal. The overall hardness is in a downward trend with the increase of heat input, and the minimum hardness is only 159 HV_0.3 at 40 J/mm. The heat input falls within the range of 28–30 J/mm, and the temperature shock resistance is at its peak.

With the rapid development of integrated circuits, glass substrates are frequently utilized for prototyping various functional electronic circuits due to their superior stability, transparency, and signal integrity. In this experiment, copper wire was printed on a glass substrate using inkjet printing, and the electronic circuit was sintered through laser irradiation with a 532 nm continuous green laser. The relationship between resistivity and microstructure was analyzed after laser sintering at different intensities, scanning speeds, and iterations. The experimental results indicate that the conductivity of the sintered lines initially increases and then decreases with an increase in laser power and scanning speed. At the same power level, multiple sintering runs at a lower scanning speed pose a risk of increased porosity leading to reduced conductivity. Conversely, when the scanning speed exceeds the optimal sintering speed, multiple sintering runs have minimal impact on porosity and conductivity without altering the power.

Composite coatings reinforced with varying mass fractions of SiC particles were successfully fabricated on 316 stainless steel substrates via laser cladding. The phase compositions, elemental distribution, microstructural characteristics, hardness, wear resistance and corrosion resistance of the composite coatings were analyzed using X-ray diffraction (XRD), scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDS), Vickers hardness testing, friction-wear testing and electrochemical methods. The coatings have no obvious pores, cracks or other defects. The phase compositions of the Hastelloy C276 coating includes γ-(Ni, Fe), Ni₂C, M₆C, M₂(C, N) and M₂₃C₆. SiC addition resulted in the formation of high-hardness phases, such as Cr₃Si and S₅C₃, with their peak intensity increasing with SiC content. The dendrites extend from the bonding zone towards the top of the coatings, and the crystal direction diffuses from the bottom to each area. Compared with the dendritic crystals formed at the bottom, the microstructure at the top is mostly equiaxed crystals and cellular crystals with smaller volume. When SiC powder particles are present around the crystals, the microstructure of the cladding layer grows acicular crystals containing Si and C. These acicular crystals tend to extend away from the residual SiC powder particles, and the grain size in this region is smaller and more densely distributed. This indicates that both melted and unmelted SiC powder particles can contribute to refining the grain structure of the cladding layer. The optimal SiC addition was determined to be 9 wt%, yielding an average microhardness of 670.1 HV_0.5, which is 3.05 times that of the substrate and 1.19 times that of the 0 wt% SiC coating. The wear resistance was significantly enhanced, reflected by a friction coefficient of 0.17 (43.59% of the substrate, 68% of 0 wt%) and a wear rate of 14.32 × 10⁻⁶ mm³N⁻¹·m⁻¹ (27.35% of the substrate, 40.74% of 0 wt%). The self-corrosion potential measured at 315 mV, with a self-corrosion current density of 6.884 × 10⁻⁶ A/cm², and the electrochemical charge-transfer resistance was approximately 25 times that of the substrate and 1.26 times that of the 0 wt%. In this work, SiC-reinforced Hastelloy-SiC composite coating was studied, which provides a new solution to improve the hardness, wear resistance and corrosion resistance of 316L stainless steel.

To address the problem of excessive ablation in conventional laser processing caused by the inhomogeneous energy distribution at the focal point, along with the inherent heterogeneity and surface irregularities of textile materials, a new method for laser printing elastic bandage fabrics was developed. We used flat top light sources, short focal field mirrors, and low power lasers instead of the Gaussian light sources, long focal field mirrors, and high-power lasers used in traditional methods. First, the sample was preheated, and the aspherical lens system was designed and simulated. Then, the physical and chemical properties of laser-processed elastic bandage fabrics were investigated. Finally, based on single-factor experiments, orthogonal experimental analysis was conducted to determine the optimal process parameters. The results show that the optimized optical path can effectively improve the uniformity of the temperature field during laser scanning and enhance focusing performance; as energy gradually accumulates, chemical bonds in polymer molecules break; when the elastic bandage fabric is in a highly elastic state, it exhibits appropriate breaking strength and color difference. The best parameters obtained from the single-factor experiment are as follows: laser power range of 25–34 W, scanning speed range of 2200–2800 mm/s, preheating temperature range of 125–200 °C. The best parameters obtained from the orthogonal experiment are as follows: laser power 28 W, scanning speed 2800 mm/s, and the preheating temperature 175 °C.