Bin Yuan

Quality inspection of steel plate welding is critical in industrial manufacturing. However, weld proximity defects often present diverse morphologies, overlapping regions, and dense distributions, posing challenges to accurate industrial defect inspection. Therefore, we propose an industrial detector based on the EP-YOLOv7. First, an Efficient Multiscale Channel Attention (EMCA) is introduced to strengthen multi-scale feature perception and improve the model’s focus on weld proximity defects. Second, the EMCA module is integrated into the Efficient Layer Aggregation Network to enhance feature fusion and defect representation. Finally, a Partial-Bottleneck Decoupling Predictor Head (P-BD Head) is designed to significantly improve localization accuracy and reduce missed detections of small targets. Experimental evaluations on both a self-built weld proximity defect dataset and a public generalization dataset show that EP-YOLOv7 achieves mAP of 85.2%/56.2% and F1 scores of 80.3%/43.3%. Meanwhile, the model size increases by only 0.6 MB (total 37.9 MB), demonstrating that the proposed approach delivers substantial accuracy gains while maintaining lightweight computational complexity, suitable for practical industrial applications.

Aluminum alloy is a material of particular significance within the domain of light weighting research. The present study investigates the effects of welding parameters on the forming and mechanical properties of static shoulder stir friction welded joints of 3 mm thick 6003 aluminum alloy. It was determined that the optimal quality of welds, characterized by their formation and absence of defects, was achieved through the utilization of static shoulder stir friction welding at speeds ranging from 3.000 to 4.000 rpm, with ω/v maintained within the range of 3 to 4. These results emphasize the importance of the welding parameters on the results.

Optimization and active control of internal gearing power honing (IGPH) process parameters for excellent and stable gear precision were carried out using the engagement theory of a conjugate curved face, the Box–Behnken design of experiments method, and the artificial immune clone selection algorithm (AICSA). Optimization and active control were carried out in four stages. In the first stage, the second-order models of tooth profile deviations were developed considering the nonlinear influence of IGPH process parameters on tooth profile deviations based on the Box–Behnken design. In the second stage, a method for solving the multi-objective optimization of the IGPH process was presented based on building the synthetic tooth profile deviation model, which considered the different weighting factors of different tooth profile deviation indexes. In the third stage, excellent gear precision was obtained by importing the ranges of synthetic tooth profile deviation and parameters into the AICSA. In the fourth stage, based on the optimized process parameters, the active control of IGPH process parameters was realized based on the constant cutting speed on the fixed position of the gear tooth surface. The total gear profile error reached a minimum value at the optimal parameters of 1270.4 rpm for spindle speed, 60 mm min⁻¹ for axis feed velocity, 2.4 µm per oscillation for radial feed velocity, and 2.4 spark-out times. The gear accuracy test results show that the total gear profile error value from the above active control method is more stable and lower than that without active control, indicating that the proposed method is effective.