Image and Feature Space Based Domain Adaptation for Vehicle Detection

The application of deep learning in the field of object detection has experienced much progress. However, due to the domain shift problem, applying an off-the-shelf detector to another domain leads to a significant performance drop. A large number of ground truth labels are required when using another domain to train models, demanding a large amount of human and financial resources. In order to avoid excessive resource requirements and performance drop caused by domain shift, this paper proposes a new domain adaptive approach to cross-domain vehicle detection. Our approach improves the cross-domain vehicle detection model from image space and feature space. We employ objectives of the generative adversarial network and cycle consistency loss for image style transfer in image space. For feature space, we align feature distributions between the source domain and the target domain to improve the detection accuracy. Experiments are carried out using the method with two different datasets, proving that this technique effectively improves the accuracy of vehicle detection in the target domain.     

磁性磨粒辅助磁针磁力研磨的应用研究

目的解决磁针磁力研磨工艺中磁针对工件表面碰撞损伤及存在研磨盲区的问题。方法在磁针中加入磁性磨粒增加磁针束的柔性,同时磁针为磁性磨粒提供研磨压力和切削力。将三相正弦交流电接入定子线圈,利用交流电的相位差产生旋转磁场,驱动混合磨料对微小复杂工件进行研磨。在混合磨料总质量不变的条件下,依次采用磁针、磁性磨粒和不同质量混合比的混合磨料进行对比试验。结果相较于单一磨料,使用混合磨料加工40 min后的工件表面形貌较好,表面粗糙度值下降幅度大,且有较大的材料去除量。当磁针与磁性磨粒的质量混合比为1∶2时,加工后的工件表面形貌最佳,无明显加工纹理和磁针碰撞痕迹,工件表面粗糙度值由原始的1.0μm下降到0.54μm左右,材料去除量为2.8 mg左右,微小沟槽内无研磨盲区。结论在电磁研磨工艺中,使用磁针和磁性磨粒质量比为1∶2的混合磨料可提高研磨效果,避免磁针的碰撞对工件表面造成损伤,磁针可将磁性磨粒挤入工件微小沟槽,无研磨盲区。     

Effects of temperature and strain rate on microstructure and mechanical properties of high chromium cast iron/low carbon steel bimetal prepared by hot diffusion-compression bonding

The objective of this study is to develop a hot diffusion-compression bonding process for cladding low carbon steel (LCS) to high chromium cast iron (HCCI) in solid-state. The influence of temperature (950–1150 °C) and strain rate (0.001–1 s⁻¹) on microstructure, hardness and bond strength of the HCCI/LCS bimetal were investigated. The interface microstructure reveals that the unbonded region can only be found for 950 °C due to lack of diffusion, while the intergrowth between the constituent metals occurred at and above 1100 °C. When bonding temperature increases to 1150 °C, a carbide-free zone was observed near the interface on the HCCI layer, and the thickness of the zone decreases with an increase of bonding strain rate. These evolutions indicate that the bond quality was improved by raising temperature and reducing strain rate due to the increase of element diffusion. The hot compression process of the bonding treatment not only changes the carbide orientation of the HCCI, but also increases the volume fraction of Cr–carbide. Based on the microstructural examinations and mechanical tests, the optimum bonding temperature and bonding strain rate are determined to be 1150 °C and 0.001 s⁻¹, respectively.     

Quantitative Analysis of Work Hardening and Dynamic Softening Behavior of low carbon alloy Steel Based on the Flow Stress

In this study, the constitutive equation and DRX(Dynamic recrystallization) model of Nuclear Pressure Vessel Material 20MnNiMo steel were established to study the work hardening and dynamic softening behavior based on the flow behavior, which was investigated by hot compression experiment at temperature of 950 °C, 1050 °C, 1150 °C and 1250 °C with strain rate of 0.01 s⁻¹, 0.1 s⁻¹ and 10 s⁻¹ on a thermo-mechanical simulator THE RMECMASTOR-Z. The critical conditions for the occurence of dynamic recrystallization were determined based on the strain hardening rate curves of 20MnNiMo steel. Then the model of volume fraction of DRX was established to analyze the DRX behavior based on flow curves. At last, the strain rate sensitivity and activation volume V^* of 20MnNiMo steel were calculated to discuss the mechanisms of work hardening and dynamic softening during the hot forming process. The results show that the volume fraction of DRX is lower with the higher value of Z (Zener–Hollomon parameter), which indicated that the DRX fraction curves can accurately predicte the DRX behavior of 20MnNiMo steel. The storage and annihilation of dislocation at off-equilibrium saturation situation is the main reason that the strain has significant effects on SRS(Strain rate sensitivity) at the low strain rate of 0.01 s⁻¹ and 0.1 s⁻¹. While, the effects of temperature on the SRS are caused by the uniformity of microstructure distribution. And the cross-slip caused by dislocation piled up which beyond the grain boundaries or obstacles is related to the low activation volume under the high Z deformation conditions. Otherwise, the coarsening of DRX grains is the main reason for the high activation volume at low Z under the same strain conditions.     

Generalized wave equation in nonlocal elasticity

The refined theory of torsional deformation of a circular shaft is proposed on the basis of the classical elasticity theory. The refined theory provides the solutions of torsional deformation of a circular shaft without ad hoc assumptions. Exact solutions, including a Saint-Venant’s solution part and a transcendental part, are obtained for a circular shaft with homogeneous boundary conditions, and the decomposed theorem is given and proved. The approximate solutions are derived for a circular shaft under reverse direction surface loading by dropping terms of high order. To illustrate the application of the theory developed, two examples are examined.     

大范围运动条件下的刚柔混合机械臂动力学分析

建立了刚-柔混合两杆机械臂的动力学方程,并进行了数值仿真,分别讨论了在大范围运动已知条件下和大范围运动未知条件下,刚-柔混合机械臂的动力学响应。在大范围运动未知条件下,又讨论了刚性机械臂和柔性机械臂的运动对系统动态响应的影响。     

基于逐步回归分析法的表面粗糙度预测

为减少渗碳钢轴的导磁性对磁粒研磨加工的影响,将磨料贮存于周围镶嵌多个磁极的圆形容器中,使被工件带走的磁性粒子能够在磁极间循环利用,虽然降低了磁性磨料在单个磁极处的自我更新作用,但是能够防止磁性磨料的流失。开展对20CrMnTi材质的轴类零件的试验,以加工时间、工件转速、磁性磨料和研磨液的质量比、磁性磨粒的粒径为自变量,工件表面粗糙度作为因变量,采用逐步回归分析建立表面粗糙度预测模型,通过试验验证预测结果的准确性。结果表明:预测模型的表面粗糙度的相对误差绝对值能够控制在7%以内,具有较好的预测能力。      

Effect of denucleating glass composition on undercooling of Fe83Ga17 alloy melts

Adopting glass fluxing combined with superheating cycling method, the undercooling and its stability of Fe₈₃Ga₁₇ alloy melts were investigated using different kinds of denucleating glass: B₂O₃, 90% NaSiCa + 10% B₂O₃ (simplified as Na–Si–Ca–Al–B) and 70% Na–Si–Ca–Al–B + 30% Na₂B₇O₄. The results showed that different glass has different denucleating mechanism. The purification of B₂O₃ glass is only a physical process, by which the stable bulk undercooling cannot be obtained during superheating–cooling cycles. While taking Na–Si–Ca–Al–B glass as purifying agent, its denucleating mechanism is a comprehensively physicochemical process. But the stability of undercooling is still undesirable because of the separation between melt and glass during cooling process in superheating cycling. A stable bulk undercooling can be obtained by physicochemical denucleating process in the case of 70% Na–Si–Ca–Al–B + 30% Na₂B₇O₄ molten glass owing to its suitable viscosity.     

相场法在金属凝固过程的应用现状

摘要：相场法是在计算材料学中发展最快的一种强大的计算方法,以基本的热力学和动力学为输入,可用于模拟和预测材料的介观尺度形貌和微观结构的演变。首先总结了相场模型的历史发展、物理基础、数学表达及其数值求解,其次分析了相场法在纯物质、二元合金、多组分系统以及定向凝固、增材制造等领域中的应用情况,最后对相场法进行了总结及展望,并指出相场法发展应趋向于超大尺度相场模拟技术,更高效算法的开发,相场模型与热力学、动力学数据库的结合,工业应用的探索以及与实验观察技术的进一步结合。     

基于Fluent的蝶阀三维流动的数值模拟及分析

结合生产实际,针对大口径蝶阀,运用商用流体计算软件FLUENT,对其不同开度情况下的流场形式进行了三维数值模拟分析。特别是根据分析结果对大口径蝶阀结构进行了相应的改进。分析表明,改进后的蝶阀性能明显提高,流阻系数明显降低,对阀门优化设计有着积极的借鉴作用。