Abstract: Aiming at the pain points of long debugging cycle, insufficient precision and poor adaptability to 3D scenes in traditional path planning of welding manipulators for new energy vehicle battery packs, this paper builds a high fidelity welding simulation environment based on Unity engine and ML Agents toolkit, introduces the DQN deep reinforcement learning algorithm, designs a multi-dimensional composite reward function, and completes the training and verification of the weld seam path planning agent. Comparative experiments show that compared with the traditional Q-Learning algorithm, the DQNalgorithm reduces the welding path error by 34.7% and improves the planning efficiency by 22%, which can effectively meet the high-precision production requirements of battery pack welding and provide a feasible solution for intelligent path planning of industrial welding.

Abstract: A partially butt-welded 6005A aluminum alloy sidewall structure of a high speed train was selected as the research object, and a thermo-mechanical sequential coupled finite element model considering the clamp constraint and release process was established. The post-weld deformation and residual stress distributions under four working conditions—free condition (FC), continuously clamped condition (CC), immediate release condition (IR), and release after cooling to 100℃ (TR)—were compared and analyzed. The results show that, under the condition of consistent welding temperature field distribution, the timing of clamp release mainly affects the post-weld mechanical response. Under all four working conditions, the out-of-plane deformation of the component is mainly negative displacement. The continuous constraint working condition has the most obvious inhibitory effect on the out-of-plane deformation after welding. In contrast, the release working conditions all show a certain degree of spring back effect after the clamp is released. The longitudinal residual stresses under all four conditions show the typical distribution characteristics of tensile stress concentration in the weld center area and gradually transforming into compressive stress away from the weld area. The peak value order is CC > IR > TR > FC. For transverse residual stress, the CC condition maintains a higher tensile stress level along the weld direction, which is significantly higher than the other working conditions. Studies have shown that the timing of clamp release essentially alters the distribution relationship between deformation release and stress retention during welding shrinkage. Appropriately extending the clamp holding time helps to achieve a better balance between post-weld deformation control and residual stress reduction.

Abstract: Bionic tribology, a new couping of biology and tribology, provides a new angle to solve an urgent problem concerning friction-wear-lubrication. This review systematically presents the basic rules of biomimetic surface design for natural biological systems which possess excellent tribological properties as inspiration. Detailed design strategies such as micro/nano- structure fabrication, surface modification and material selection are introduced in terms of the latest progress in the manufacturing technologies. Typical applications in aerospace, automotive, biomedical as well as marine engineering are discussed with the focus on the improvement of physical attributes and technological developments. Current bottlenecks in manufacturing scalability, durability under harsh conditions and theoretical modeling are discussed to compare and contrast the proposals together with outlooks for future research and perspectives like machine learning-assisted design and multifunctional adaptive surfaces. It is hoped that this comprehensive review on design, fabrication and applications of bionic tribological surfaces will facilitate the sustainable engineering.

Abstract: Flexible piezoresistive sensors with high sensitivity and breathability are critical for next-generation wearable healthcare monitoring. However, achieving a balance between a wide detection range and long-term structural stability remains a significant challenge. Here, we report a breathable, high-performance piezoresistive sensor based on a synergistic MXene/PEDOT:PSS hybrid coating deposited on an electrospun thermoplastic polyurethane (TPU) nanofiber scaffold. An optimized MXene:PEDOT:PSS ratio of 2:1 results in a low sheet resistance of 12.4 Ω/sq while maintaining high air permeability (25.4 mm/s). The sensor exhibits a high gauge factor of 186, a rapid response time of 65 ms, and a low detection limit of 0.1% strain. The “brick-and-mortar” hybrid architecture, in which PEDOT:PSS chains function as conductive bridges across micro-cracks in the MXene network, provides enhanced electrical stability and mechanical durability. As a result, the sensor maintains stable performance over 5,000 loading cycles and demonstrates washability through 10 domestic laundering cycles (ISO 6330). Practical demonstrations confirm the sensor’s capability to detect subtle physiological signals, including the percussion (P), tidal (T), and dicrotic (D) waves in radial pulse monitoring, as well as complex speech phonation patterns and biomechanical gait cycles. These results highlight a promising strategy for designing breathable, durable, and multifunctional e-textile sensors for wearable health monitoring applications.