To investigate the impact of asymmetric wheel wear on vehicle operation, a finite element model of wheel-rail contact was established and solved to account for the influence of asymmetric tread wear on the wheel profile. Integrating measured wheel tread data with a wheel wear simulation model based on the Archard wear model, this study examined the wheel-rail contact relationship, tread wear distribution, and equivalent conicity under asymmetric wheel wear conditions. Research findings reveal significant changes in the positions of wheel-rail contact points on both sides under asymmetric wear conditions. Correspondingly, the equivalent taper increases markedly under asymmetric wear. Furthermore, when the lateral displacement of the wheelset exceeds 8.5 mm, asymmetric wear exhibits a pronounced upward trend. Subsequent dynamic simulations revealed that asymmetric wheel wear induces relatively higher stresses on one side, exacerbating uneven wear across both sides. This alters the wheel-rail contact relationship, thereby degrading train dynamic performance. Finally, a multi-objective optimization of the suspension parameters for the EMU was performed using Isight and Simpack software based on the NSGA-II algorithm to find the optimal solution. The results indicate that after optimization, the wear difference between the two sides of the wheel was reduced by 2.1%, and the vehicle's dynamic performance was also improved.

Accurately acquiring cutting forces and temperatures is essential for understanding cutting mechanisms and optimizing machining processes. However, predicting these forces and temperatures becomes extremely challenging when the combined effects of tool edge geometry and cutting parameters are considered. Traditional empirical equations and functional models exhibit limitations regarding prediction accuracy and computational efficiency. To address this issue, this study proposed a back-propagation neural network model optimized by the Levenberg-Marquardt algorithm (LM-BP), utilizing physical simulation data as training samples. The model was validated through machining experiments, demonstrating higher prediction accuracy and faster convergence speed. The prediction errors for cutting forces and temperatures were less than 4% relative to the experimental values. The model effectively predicted the combined effect of cutting edge radius, form factor (K), feed rate and cutting speed on cutting forces and temperatures. Based on the prediction results, the rational matching relationship between edge geometry and cutting parameters was investigated. The findings indicate that the primary cutting force and radial force are minimized at a feed-to-cutting edge radius ratio (R_fr) of 5, which also corresponds to a lower cutting temperature. According to calculations based on the Usui wear model, the tool wear rate is lowest when K is 0.9 and R_fr is 5. The proposed prediction model provides a basis for designing cutting tool edge geometries and the optimal matching with cutting parameters in precision machining.

Cylinder liner–piston ring (CLPR) tribo-pairs are essential components in internal combustion engines (ICEs), where the compatibility between piston ring coatings and cylinder liner materials strongly affects friction, wear, and engine reliability. Although piston ring coatings have demonstrated excellent tribological performance in previous studies, most investigations have focused on coating properties themselves, while the matching behavior between steel cylinder liners and multiple commercial piston ring coatings under different lubrication conditions remains insufficiently understood. To address this issue, this study systematically evaluates the tribological compatibility of two cylinder liner materials—as-cast bainite and 45# steel—with four commercial piston ring coatings (PVD, DLC, CKS, and CDC). Reciprocating friction tests were conducted using an SRV tribometer under dry friction, starved lubrication, and flooded lubrication conditions. Friction coefficient, wear mass, and scuffing resistance time were measured to assess the matching performance of different material combinations. The results show that the 45# steel liner paired with DLC-coated rings consistently exhibits lower friction, reduced wear, and longer scuffing resistance time than the other combinations, particularly under high-load and low-viscosity oil conditions. SEM and EDS analyses of the worn surfaces indicate milder adhesion-related wear, shallower wear scars, and fewer transfer features for the steel liner–DLC ring pair compared with the steel liner–PVD ring pair. These findings highlight the importance of material matching in determining the tribological performance of CLPR systems under different lubrication states.

Plastics are widely used owing to their light weight and durability; however, excessive production and disposal have caused serious environmental problems. As an alternative material, 100% self-bonded bamboo plates made from “fine bamboo fibers” extracted using a machining center have been developed. However, variations in the fiber shape, induced by forced vibration during bamboo-culm cutting, affect the product quality. In bamboo-culm end milling, the cross section of the culm is nearly elliptical; therefore, the vibration modes and natural frequencies vary depending on the cutting position. This study proposes a spindle-speed determination method using a time–spindle speed–vibration acceleration color map to derive a dynamic spindle-speed condition that avoids the resonance regions. Furthermore, experimental results, FEM analysis, and a reduced two-degrees-of-freedom dynamic model were compared and examined as data sources for constructing the color map, and the validity and applicability of each approach were evaluated. Based on these investigations, the derived dynamic spindle-speed condition obtained from the dynamic model was applied to the machining process, demonstrating the avoidance of resonance regions and significant reduction in the vibration acceleration. Finally, self-bonded bamboo plates fabricated from fine bamboo fibers produced under the dynamic spindle-speed condition exhibited improved uniformity in surface coloration and enhanced tensile strength.

Polycrystalline diamond compact (PDC) thrust bearings used in turbodrills are prone to the PDC insert edge fractures and surface material spalling during actual operations. Meanwhile, excessive wear on individual bearings is a frequently observed issue in series bearing assemblies, preventing the entire assembly from meeting its service life requirements. To address the unclear load distribution and the load-sharing performance of PDC thrust bearings, a finite element model was established for mechanical analysis, and the load capacity and load-sharing capability were investigated. The research results show that the PDC thrust bearing exhibits a certain degree of eccentric loading under both single-stage and series loading conditions. During engagement, PDC inserts operate under cyclic alternating stress, with the stress consistently concentrating near their edges. Both the extreme value and standard deviation of the maximum equivalent stress in the PDC insert are inversely proportional to the insert diameter and number. Appropriately increasing the PDC insert diameter and number can effectively enhance the load capacity of PDC thrust bearings while also ensuring smoother operation. When PDC thrust bearings are loaded in series, the load at each stage is unevenly distributed. According to the load distribution law of multi-stage bearing assemblies, applying variable stiffness support to each stage can effectively alleviate the uneven load distribution. For a seven-stage bearing set, the maximum overload ratio is reduced to 10.4%, which is 33.2% lower than that under uniform stiffness. The analysis results can provide valuable guidance for the optimal design of PDC thrust bearings and the design of turbodrill support sections.

Carbon fiber-reinforced polymer (CFRP) composites combined with titanium alloys (TC4) and similar metallic alloys are extensively employed in modern aerospace structures. Owing to the distinct physical and mechanical characteristics of these two categories of materials, adaptive machining with variable process parameters has been regarded as an effective strategy to enhance the quality and dimensional accuracy of machined holes. Nevertheless, most aerospace components feature complex configurations with variable thicknesses and curved surfaces. In practice, notable deviations often exist between the nominal layer thicknesses and manually measured values, which increases the difficulty in recognizing the material transition interface and restricts the real-time adjustment of process parameters during hole-making operations. Moreover, in this work, an online interface recognition algorithm based on a dynamic threshold criterion is proposed, in which the current signal collected from the machining equipment is utilized as the detection basis. With the aid of a self-developed portable helical milling system, experimental investigations involving variable-parameter machining are carried out on CFRP/TC4 stacked assemblies. The experimental results demonstrate that the developed algorithm yields high identification precision, with the detection errors controlled within 0.08 mm for the entry interface, 0.14 mm for the interlaminar interface, and 0.2 mm for the exit interface, respectively.

Text-to-image generation technology based on diffusion models has achieved remarkable results in the fields of general visual generation and digital product design. However, in complex industrial machine tool design scenarios, the lack of domain-specific prior knowledge and engineering rule constraints leads to poor structural compliance of generated results, insufficient cross-modal semantic alignment, and a substantial increase in computational costs and memory overhead during edge deployment of the model. Moreover, the design iteration speed fails to meet the requirements of personalization and speed compatibility. To address these issues, this paper proposes a rule-compliant structure-aware diffusion framework (B-MIG-SCA). This framework integrates three core components: a bidirectional mutual-information-gated mechanism that dynamically balances linguistic and structural priors, a differentiable industrial specification constraint module that converts 3D assembly and tolerance rules into 2D optimization losses, and a low-rank key-value caching strategy. Extensive comparative experiments demonstrate that under the GPU environment, the proposed model achieves a low Fréchet Inception Distance (FID) of 15.3, a CLIP score of 0.89, and a Layout-IoU of 0.82. In CPU lightweight deployment, it not only maintains a stable FID of 19.2 but also achieves a 4.6-fold significant reduction in memory consumption and a 3.1-fold improvement in inference throughput compared with baseline large models. Furthermore, the model retains a structural topology accuracy of 90.2% when facing defective text interference and achieves a high score of 4.45/5.00 in the expert subjective evaluation of structural rationality. This framework effectively balances generation quality, physical feasibility, and deployment efficiency, providing an efficient practical solution for intelligent industrial generative design.

Japan is facing an aging population, and the number of agricultural workers continues to decrease year by year. Although the Japanese government is promoting automation in agriculture, many fruit farming tasks, such as harvesting and cultivation, still require delicate and precise manual handling to maintain fruit quality. As a result, fruit farmers often perform long periods of overhead work while keeping their arms raised, which places a significant physical burden on the upper limbs and may increase the risk of musculoskeletal disorders. To reduce the physical burden, Koay et al. proposed a non-powered arm support suit equipped with Drum Type Mechanical Brakes (DTMBs). By pressing the levers mounted on the suit, the DTMBs can be engaged to support the user’s arms at elevated postures, and the support can be maintained even after the levers are released. Releasing the brake is also easy. Because the suit consists only of passive mechanical components, it operates without electricity, provides inherent safety, and avoids battery-related limitations. Notably, while the locking mechanism used in the DTMB relies on friction, it is a mechanism that retains the advantages of friction-based mechanisms while eliminating their disadvantages. However, the design method for the key dimensions of the DTMB's locking mechanism has never been clarified. Therefore, this paper presents the design method for the key dimensions of the DTMB's locking mechanism. First, the features of the non-powered arm support suit are introduced. Second, the structure and mechanism of the DTMB are described. Third, the design method for the key dimensions of the DTMB's locking mechanism is explained. Finally, experimental results are presented to validate the DTMB’s locking mechanism developed using the design method and confirm that it functions as intended.

Particle swarm optimization (PSO) has been widely applied to constrained optimization problems (COPs) due to its simplicity and effective convergence capability. In our previous work, we proposed independent two-group particle swarm optimization (I2GPSO) for COPs. I2GPSO employs a dual-population framework to balance exploration and exploitation and has demonstrated competitive performance compared with state-of-the-art PSO variants. However, I2GPSO uses the same PSO parameter settings and applies search operator uniformly across all particles, regardless of individual particle performance, which may limit its ability to effectively balance exploration and exploitation. Furthermore, its constraint-handling strategies require problem-dependent parameter tuning and may bias the search toward feasible regions, increasing the risk of premature convergence. This paper presents a novel PSO algorithm for COPs, termed ranking-based adaptive independent two-group particle swarm optimization (RA-I2GPSO). Building upon I2GPSO, RA-I2GPSO introduces four key extensions: particle ranking based on Deb’s feasibility rule, ranking-based PSO parameter control to adjust individual particle behavior, ranking-based search-operator control to further enhance global exploration and local exploitation, and redesigned constraint-handling strategies. The performance of RA-I2GPSO was evaluated on thirteen G-series benchmark problems and three engineering design problems. Experimental results demonstrated that RA-I2GPSO achieved superior solution quality and search efficiency compared with several state-of-the-art evolutionary algorithms.