Recently, researchers have been studying methods to introduce deep learning into automated optical inspection (AOI) systems to reduce labor costs. However, the integration of deep learning in the industry may encounter major challenges such as sample imbalance (defective products that only account for a small proportion). Therefore, in this study, an anomaly detection neural network, dual auto-encoder generative adversarial network (DAGAN), was developed to solve the problem of sample imbalance. With skip-connection and dual auto-encoder architecture, the proposed method exhibited excellent image reconstruction ability and training stability. Three datasets, namely public industrial detection training set, MVTec AD, with mobile phone screen glass and wood defect detection datasets, were used to verify the inspection ability of DAGAN. In addition, training with a limited amount of data was proposed to verify its detection ability. The results demonstrated that the areas under the curve (AUCs) of DAGAN were better than previous generative adversarial network-based anomaly detection models in 13 out of 17 categories in these datasets, especially in categories with high variability or noise. The maximum AUC improvement was 0.250 (toothbrush). Moreover, the proposed method exhibited better detection ability than the U-Net auto-encoder, which indicates the function of discriminator in this application. Furthermore, the proposed method had a high level of AUCs when using only a small amount of training data. DAGAN can significantly reduce the time and cost of collecting and labeling data when it is applied to industrial detection.

Silicon carbide (SiC) has significant potential as a third-generation semiconductor material due to its exceptional thermal and electronic properties, yet its high hardness and brittleness make processing costly and complex. This study introduces ultraviolet laser ablation as a method for direct SiC material removal, investigating the effects of varying scanning speeds on surface composition, hardness, and ablation depth. The results indicate optimal processing speeds for the Si and C faces at 200 mm/s and 100 mm/s, respectively. Ablation depth is linearly correlated with laser repetitions, achieving a 25% improvement in removal efficiency at 100 mm/s on the C face compared to higher speeds. A composition analysis shows that the Si and C faces of SiC exhibit consistent ratios of Si, O, and C both before and after ablation. Post-ablation, the proportion of Si and C decreases with an increased presence of oxygen. At scanning speeds below 200 mm/s, the variation in speed has minimal effect on the compositional ratios, indicating a stable elemental distribution across the surface despite differences in processing speed. Hardness testing indicates an initial hardness of 13,896 MPa for the C face, higher than that of the Si face, with both surfaces experiencing a drop to less than 1% of their original hardness (below 50 MPa) after ablation. Lattice structure analysis shows Moissanite-5H SiC and cubic silicon formation on the Si face, while the C face retains partial SiC structure. This study found that when laser parameters are used to process SiC, the processing parameters required on both sides are different and provide important reference information for future industrial processing applications to shorten the time and process cost of SiC surface thinning.