| Title | Keywords | ||
|---|---|---|---|
| Author | Authorship | ||
| Corresponding Author | Funds | ||
| DOI | Column | ||
| Summary | |||
| Timeframe | - | ||
| Title | Keywords | ||
|---|---|---|---|
| Author | Authorship | ||
| Corresponding Author | Funds | ||
| DOI | Column | ||
| Summary | |||
| Timeframe | - | ||
2A97-T3 Al-Li alloy sheets with thickness of 2.8 mm were selected as the test materials. Filler wire welding on the inner corner of T-joints was realized via wire-filled stir welding process, and sound defect-free T-joints with satisfactory forming quality were obtained. Combined with visual inspection, ultrasonic testing, industrial CT testing and cross-sectional metallography, the effects of filler wire profile, assembly gap and process parameters on the welding forming quality of T-joints were investigated. The results show that compared with circle filler wires, square filler wires deliver superior weld forming quality. Defects initiate in T-joints when the gap between skin and stringer plate reaches approximately 1.09 mm. Defect size rises with increasing gap value, defect morphology evolves from tiny voids to massive insufficient filling, and defect locations extend gradually along the height direction of the stringer. Sound defect-free T-joints with satisfactory forming quality can be produced using square filler wires under the rotational speed of 2000 r·min-1and the welding speed ranging from 50 to 100 mm·min-1.
The accuracy of rolling force prediction directly affects the setting of rolling schedules and the effectiveness of online control, serving as a crucial guarantee for stable and efficient cold rolling production. To improve rolling force prediction accuracy, a mechanism-data hybrid-driven cross-process cold rolling force prediction method was proposed. First, industrial big data platform was utilized to obtain hot and cold rolling data, and an integrated extreme learning machine(ELM) was employed for regression modeling of deformation resistance to correct the traditional rolling force calculation model. Then, the theoretical calculation values from the corrected model were used as inputs for the extreme gradient boosting(XGBoost) model, and key hyperparameters were globally optimized via genetic algorithm(GA) to enhance the model′s prediction accuracy and stability under industrial conditions. The results demonstrate that the proposed GA-XGBoost hybrid model improves the root mean square error eRMSE by 4.23% and 14.93% compared to the GA-XGBoost and XGBoost models on the test set, with R2 of 0.9841 and average absolute error eMAE reduces to 54.36 kN. The model shows significant improvement in prediction accuracy and exhibits good generalization capability, providing valuable insights for high-precision cold rolling force prediction.
Static mechanical property tests were conducted on HC1000/1470DP dual-phase steel, and five types of failure specimens with different configurations were designed based on the GISSMO failure criterion. A combined hardening model was established using the Voce++ and Hockett-Sherby equations. The undetermined fitting coefficients of the model were obtained by combining failure tests and numerical simulations, and the stress triaxiality as well as the curves and corresponding parameters of the GISSMO fracture failure criterion for the five specimens were acquired. The accuracy of the established characterization model was verified through correlation analysis between static three-point bending test and simulations of hat-shaped beam.The analysis results show that the average maximum error of the key parameters for the fracture failure characterization model based on the GISSMO failure criterion is 7.05%. The load-displacement curves obtained from the three-point bending tests and simulations of hat-shaped beam present consistent variation trends. The peak load average error is approximately 2.34% and the corresponding displacement average error is 4.19%. It indicates that the established material fracture failure model possesses high characterization accuracy.
To accurately predict the fracture properties of aluminum alloy tailor-welded blanks(TWBs) during spinning, the spinnability experiments and fracture prediction research on 2219 aluminum alloy friction stir welded(FSW) TWBs were conducted. The weld zone was characterized and partitioned through metallographic analysis, hardness testing and tensile experiments, based on which a partitioned material model was established. A study on the shear-spin spinnability experiment and fracture prediction of TWBs based on the ductile fracture criterion was completed. The results show that the introduction of welds reduces the spinnability of 2219 aluminum alloy TWBs, with fractures occurring at the inner surface near the weld. Among the three models, the Lou criteria demonstrates superior accuracy in predicting fracture location compared with C-L and Oyane criteria, with a prediction error of only 14.4% for the maximum thinning ratio of 2219 aluminum alloy FSW TWBs. These findings verify the feasibility of the property-partitioned criterion modeling approach and provide important technical support for the integral manufacturing of large rocket tank domes using TWBs.
Taking the drawing die for automotive panel as research object, with the adoption of CAE numerical simulation technology, accurate numerical simulation of the drawing die service process was realized through the establishment of a precise material model and reasonable parameter configuration, and a risk region identification method was applied to achieve the lightweight design of the die. A risk coefficient model that considered the real-time stress state and material real-performance of the die during service was proposed. The strength parameters of die materials with different strain rates were obtained by referring to material performance data. The secondarily developed model was embedded into simulation software to calculate the risk coefficient of each position on the die and identify low-risk and high-risk regions of the die. Lightweight treatment was conducted on the low-risk regions, and the secondary simulation under identical service conditions shows that the risk coefficients of all die positions after lightweight optimization are below the preset threshold, which verifies the feasibility of the lightweight scheme. Field tests on the physical die were finally performed, and the test results are consistent with the simulation results.
To elucidate the distribution patterns of residual stress under different experimental conditions in the microscopic evolution, residual stresses at different positions were measured and microstructures of tested specimens were characterized based on the rough machining and finish machining milling experiments for the edge-wrapped simplified forging billet of TC4 titanium alloy before and after annealing. The effects of initial residual stress and milling parameters on the milling residual stress were investigated. The results indicate that residual stress in the feed direction on the outer surface of unannealed forgings after rough milling is 1.5 times greater than that in annealed forgings, initial residual stress plays a dominant role in alleviating tensile effects during rough milling.The initial residual stress inside the forging dominates during rough milling, while it exerts a minor effect on finish milling. However, the cumulative effect of milling passes becomes predominant in finishing, leading to higher residual stress. The density of geometrically necessary dislocations(GNDs) after finish maching is increased by 48.7% compared to rough maching, indicating a stronger plastic strain gradient.
Based on the coupled Eulerian-Lagrangion(CEL) method, a finite element simulation model of friction stir welding for T-joints of dissimilar alloy plates of AA7055 alloy and AA2195 alloy was innovatively established, and the dynamic temperature field during the entire welding process was simulated and analyzed. The results show that under the process combination parameters of w=400 r·min-1, v=60 mm·min-1, and h=0.21 mm, the high-temperature area is concentrated around the shaft shoulder, and the temperature on the advancing side is higher than that on the receding side by approximately 14.6 ℃. The temperature fields of the characteristic points in the central area of the weld and the different paths in the gradient direction of the thin plate all show an asymmetric M-shaped bimodal distribution. The characteristic points far from the weld area are symmetrically distributed in an inverted V shape. The temperature of the workpiece along different paths(Z1, Z2, Z3) in the Y direction of the cross-section shows a nonlinear attenuation towards the base metal area, and the temperature in the weld core area is the highest, with the temperature gradient in the heat-affected zone being the most significant, which conforms to the thermally mechanical coupling mechanism.
Addressing the issues of low prediction accuracy in traditional rolling force mechanism models and the limitations of traditional single optimization algorithms in terms of optimization scope and accuracy, a fusion model for hot continuous rolled force prediction based on a two-layer optimization algorithm was proposed. Firstly, the LightGBM algorithm was utilized to rank the correlation of feature variables, and the top 20 features based on correlation ranking were selected as model inputs. Secondly, a BP neural network based on two-layer optimization using an improved sparrow algorithm and Bayesian algorithm was established, as well as a deep belief network optimized based on an adaptive attention mechanism and Bayesian algorithm. Finally, the outputs of the two models were weighted using a dynamic weight fusion method to obtain the final prediction value. Experimental results show that the model has an R2 value of 0.992 and an average absolute error of only 0.061, indicating high accuracy of the fused model.
AZ31 magnesium alloy semi-solid billets were subjected to asynchronous rolling below the solidus temperature to investigate the effects of rolling temperature on microstructure and mechanical properties, as well as the deformation and strengthening-toughening mechanisms. The results show that increasing temperature enhances solid particle deformation, promotes dynamic recrystallization, and refines grains, leading to improved strength and ductility; however, excessive temperatures lead to abnormal grain coarsening accompanied by coalescence and growth of recrystallized grains, which degrades the homogeneity of microstructure. The liquid-phase suppresses or delays the deformation of adjacent solid particles.After optimization, the tensile strength of 270.68 MPa, the yield strength of 219.92 MPa, and the elongation of 12.84% for the semi-solid sheet asynchronous rolled at 400 ℃ are obtained. The deformation mechanism of semi-solid billet is dominated by recrystallization of solid-phase particles, while the liquid-phase exerts a suppressive or delaying effect on deformation of adjacent solid particles; the strengthening-toughening mechanism of asynchronous rolling is dislocation strengthening and recrystallization fine-grain strengthening of solid-phase particles.
To enhance the interfacial bonding quality and overall mechanical properties of AA1060/AA7075/AA1060 laminated composite sheets, multi-layer composite sheets through three passes were successfully fabricated by room-temperature accumulative roll bonding(ARB). The effects of ARB passes on the mechanical properties, interfacial bonding characteristics, microstructure and texture evolution of the multi-layer composite sheets were systematically investigated using microhardness testing, universal tensile testing, optical microscopy, scanning electron microscopy(SEM) and electron backscatter diffraction(EBSD). The results indicate that significant differences in strength and ductility between the AA1060 and AA7075 alloy lead to inhomogeneous plastic flow during rolling, thereby causing an uneven hardness distribution along the thickness direction. The composite sheets achieve the optimal combination of mechanical properties after two passes, with yield strength and ultimate tensile strength reaching 243 and 268 MPa, respectively. As the ARB pass increases, the interfacial bonding quality progressively is improved, exhibiting a wavy morphology induced by non-uniform deformation. Both grain size and aspect ratio continuously decrease, and after three passes, the average grain sizes in the central AA1060 and AA7075 layers are refined to 0.9 μm, with the grains tending to become equiaxed. The proportion of high-angle grain boundaries in the AA1060 layer increases at a higher rate compared to the AA7075 layer, and the intensity of the cube texture first decreases and then increases. In the three passes sample, a strong rotated Goss texture and a weakly dispersed rotated cube texture form in the AA1060 layer, while such texture components are not distinctly observed in the AA7075 layer.