Hu Wu

Three-dimensional (3D) braided composites have considerable application potential in aerospace, defense, transportation, and marine engineering owing to their excellent structural integrity and mechanical properties. However, during the 3D braiding process, yarns are susceptible to fluctuations in creel tension, wear of yarn-guiding components, and variations in process parameters, causing fiber anomalies such as fiber breakage and fuzzing. Owing to their random and abrupt occurrence, these anomalies are difficult to identify accurately in real time using conventional inspection methods, thereby affecting process stability and potentially compromising preform quality. To address these challenges, this study proposes YOLOv13-SFC, a deep-learning-based fiber anomaly detection model for the real-time detection of fiber breakage and fuzzing during the 3D braiding of composite preforms. A DySample-based adaptive upsampling mechanism is introduced into the neck to enhance adaptive sampling during multi-scale feature fusion. A DSC3k2-FSAS module is proposed to synergistically integrate local convolutional representations with frequency-domain feature interactions within local patches, thereby enhancing the discriminative representation capability of the network for fine-grained textures, small anomalous regions, and complex backgrounds. Furthermore, a lightweight detection head, termed Detect-DyHead, is developed to improve the dynamic modeling and prediction of anomalies of varying scales and morphologies. Experimental results show that YOLOv13-SFC achieves an mAP50 of 92.1%, 2.7 percentage points higher than the YOLOv13n baseline. Meanwhile, the parameter count and computational cost are reduced by 8.03% and 22.58%, respectively, while the model-level inference speed reaches 49.8 frames per second. Based on the proposed model, a vision-based detection platform integrating YOLOv13-SFC with a UR5 collaborative robot is developed for online monitoring during the 3D braiding process. The end-to-end processing latency of the developed platform is 101–126 ms. Continuous-braiding experiments further demonstrate that, at a preform linear traction speed of 300 mm min ⁻¹ , the developed system is capable of online monitoring of fiber anomalies.

The three-dimensional weaving process for composite preforms is widely used in aerospace, national defense, and high-end equipment manufacturing, where mechanical properties depend strongly on fiber orientation and structural uniformity. During weaving, fibers are susceptible to abrasion, fracture, and knotting caused by bending, interlacing, and friction. These defects are typically subtle and random, making traditional inspection methods inadequate for real-time, non-destructive monitoring and ultimately degrading performance and mechanical properties. This paper presents a deep learning approach for the detection of abnormal fiber states. The BasicBlock network is restructured using a SAConv module to improve feature extraction. The integration of a RepC3-CBAM module into the ResNet18 backbone facilitates the enhanced capture of local details pertaining to small-scale fiber anomalies and improves attention to critical regions, all while preserving a lightweight network architecture. An SCConv module is incorporated into the detection head to facilitate lightweight feature reconstruction and enhance multi-scale feature representation. A robotic vision inspection platform, AUBO-i5, is developed using the proposed method, facilitating real-time monitoring of fiber anomalies in the weaving process. Experimental results indicate that the proposed model enhances mAP50 by 7.2%, decreases parameter count by 23.7%, and reduces FLOPs by 27.8%, while increasing detection speed from 74 frames per second (f/s) to 183 f s ⁻¹ . The results indicate that the proposed approach attains an optimal balance between accuracy and efficiency. The model’s effectiveness in complex industrial weaving environments is validated, offering reliable technical support for enhancing the structural uniformity and mechanical properties of three-dimensional woven fiber preforms.

Research on clean production in automotive painting processes is a core component of achieving green manufacturing, addressing environmental regulatory challenges, and advancing sustainable development in the automotive industry by reducing volatile organic compound (VOC) emissions, optimizing resource utilization, and minimizing energy consumption. To reduce pollutants generated by automotive painting processes and improve coating efficiency, this study proposes a clean production method for automotive body painting based on an improved whale optimization algorithm from the perspective of “low-carbon consumption and emission-reduced production”. A multi-level, multi-objective decision-making model is developed by integrating three dimensions of clean production: material flow (optimizing material costs), energy flow (minimizing painting energy consumption), and environmental emission flow (reducing carbon emissions and processing time). The whale optimization algorithm is enhanced through three key modifications: the incorporation of nonlinear convergence factors, elite opposition-based learning, and dynamic parameter self-adaptation, which are then applied to optimize the automotive painting model. Experimental validation using the painting processes of TJ Corporation’s New Energy Vehicles (NEVs) demonstrates the superiority of the proposed algorithm over the MHWOA, WOA-RBF, and WOA-VMD. Results show that the method achieves a 42.1% increase in coating production efficiency, over 98% exhaust gas purification rate, 18.2% average energy-saving improvement, and 17.9% reduction in manufacturing costs. This green transformation of low-carbon emission-reduction infrastructure in painting processes delivers significant economic and social benefits, positioning it as a sustainable solution for the automotive industry.

In the task of surface defect detection for ceramic materials, identifying minute defects has always posed a challenge due to the difficulty of balancing detection accuracy with computational cost. To address this issue, this paper proposes an enhanced algorithm based on RT-DETR, named RECS-DETR (RDNet-EA Attention-CARAFE-SCFF-RTDETR). First, a lightweight RD feature extraction network is innovatively designed to replace the backbone network of RT-DETR, and Efficient Additive Attention is introduced to substitute the self-attention mechanism in AIFI, thereby reducing computational costs. Next, the CARAFA upsampling module is employed to replace the Upsample module in RT-DETR, which better aggregates information and captures features. Subsequently, a novel structure called SCFF is designed specifically for tiny object detection. The SCFF small object detection optimization module introduces the S2 layer as the small object detection layer and constructs a new feature fusion pyramid, further enhancing the model's sensitivity to minute defects. Finally, a novel composite loss optimization strategy, NWD-Inner-Wise-IoU, is implemented, significantly accelerating model convergence and enhancing the detection of small objects. Experimental results demonstrate that, compared to the ResNet18-based RT-DETR algorithm, the proposed RECS-DETR achieves a 4.4 percentage point improvement in mean average precision on a ceramic surface dataset characterized by tiny defects with black spots. Additionally, it reduces the number of parameters by 45%, decreases FLOPs by 35.8%, and increases detection speed by 17%. The significantly improved performance effectively meets the requirements for detecting minute defects on ceramic surfaces in industrial production.

To reduce pollutants generated by automotive painting processes and improve coating efficiency, this study proposes a clean production method for automotive body painting based on an improved whale optimization algorithm from the perspective of "low-carbon consumption and emission-reduced production." A multi-level, multi-objective decision-making model is developed by integrating three dimensions of clean production: material flow (optimizing material costs), energy flow (minimizing painting energy consumption), and environmental emission flow (reducing carbon emissions and processing time). The whale optimization algorithm is enhanced through three key modifications: the incorporation of nonlinear convergence factors, elite opposition-based learning, and dynamic parameter self-adaptation, which are then applied to optimize the automotive painting model. Experimental validation using the painting processes of TJ Corporation's New Energy Vehicles (NEVs) demonstrates the superiority of the proposed algorithm over MHWOA, WOA-RBF, and WOA-VMD. Results show that the method achieves a 42.1% increase in coating production efficiency, over 98% exhaust gas purification rate, 18.2% average energy-saving improvement, and 17.9% reduction in manufacturing costs. This green transformation of low-carbon emission-reduction infrastructure in painting processes delivers significant economic and social benefits, positioning it as a sustainable solution for the automotive industry.

The automobile body painting production system exists in the backward management concept of focusing on production and ignoring environmental protection, focusing on traditional technology and unwilling to innovate, etc., high production cost and environmental protection standards, which have had a negative impact on the industry’s long-term growth. Based on the value stream mapping (VSM) analysis technology, the current situation of the value stream of the body spraying process of TJ Automobile Manufacturing Company is analyzed, the problem of waste in the production process of automobile body painting system affected by Eliminate Combine Rearrange Simplify (ECRS)a four-step problem-solving technique was studied and analyzed, a new method for the application of value stream graph analysis technology in the digital transformation environment is proposed, and a suitable production improvement scheme for the body painting system is formulated by using the new method, which increases the value-added operation time of the painting production process by 9.5% and reduces the non-value-added operation time by 12.5%. The number of operators was reduced by 18.8%, the value-added ratio was increased accordingly, the scrap rate was reduced by 8%, and the exhaust gas purification rate was increased by 66%, which brought good economic and social benefits to the enterprise and provided a successful case for the application of lean production VSM in the digital transformation environment. The paper extends knowledge in the area of digital transformation and lean production by helping us to establish and explain the relationship between lean production management methods and the promotion of new energy vehicle body coating performance from a dual perspective of digital transformation and VSM. None of the previous studies have considered the combined application of the studied digital transformation and VSM lean management approaches.

The efficient development of electric vehicles is essential to drive society towards sustainable development. Designing a lightweight frame is a key strategy to improve the economy and environment, increase energy efficiency, and reduce carbon emissions. Taking an automatic loading and unloading mixer truck as the research object, a force analysis of its frame was conducted under six typical working conditions. A size optimization method based on a hybrid model of the Kriging model and the analytic hierarchy process (AHP) is proposed. An approximate model of the mass and maximum stress of the frame was established using the Kriging model, and the Kriging model was optimized by using the multi-objective genetic optimization algorithm and the AHP method. Meanwhile, topology optimization was introduced to improve the structural performance of the frame and reduce its weight. The optimization results show that the overall weight of the frame is reduced by 11.96% compared to the pre-optimization period, though it still meets the material performance specifications. By comparing the iterative curves of the single Kriging model with those of the AHP model, it can be seen that the initial optimization efficiency of the hybrid model is about twice as much as that of the AHP model, and the final optimization result is improved by about 3.6% compared with the Kriging model. This validates the hybrid model as an effective tool for the multi-objective optimization of electric vehicle frames, providing more efficient and accurate optimization results for frame design.

In industrial robot welding heat exchanger plate route planning, there are drawbacks to using the traditional ant colony algorithm (ACA), including poor search efficiency and a propensity to trend toward the local optimum. To address the above issues, first of all, the article introduces the improved pheromone volatilization factor, which is adjustable based on iteration times in the ACA. Secondly, it combines the new perturbation strategy and the cross‐mutation operation, proposes an improved genetic algorithm, and fuses it into the ACA, which increases the diversity of the ACA’s path searching and improves the ability of the local and global search. Finally, the improved ACA mechanism is verified by experiments and compared with six existing path planning algorithms (including three ACA variant algorithms and three mainstream algorithms). The experimental results show that the IACAG algorithm has excellent performance in welding heat exchanger path planning and can achieve lower calculation time and iteration times under the conditions of optimal solution and zero deviation, showing comprehensive advantages in solution quality, stability, and efficiency.