基于SCAE⁃ACGAN的直升机行星齿轮裂纹故障诊断

直升机行星传动轮系结构复杂、工况多变,其振动信号受工况影响大,在故障样本较少的情况下导致行星齿轮箱故障诊断准确率不高,早期故障诊断困难。针对上述问题,提出将堆栈收缩自动编码网络(stacked contractive autoencoder,简称SCAE)与辅助分类生成式对抗网络(auxiliary classifier generative adversarial networks,简称ACGAN)相结合的SCAE-ACGAN故障诊断方法。ACGAN的生成器产生与真实样本具有类似分布的生成样本,扩展训练样本集,并与真实样本一起输入至判别器进行训练。ACGAN采用SCAE作为判别器,利用SCAE良好的抗数据波动能力,从扩展样本集中挖掘出有效的深度特征,并实现样本的真伪与类别的判定。ACGAN的判别器和生成器在对抗学习训练机制下交替优化,提高方法的样本生成质量与故障判定能力。将SCAE-ACGAN应用于直升机行星轮裂纹故障诊断,结果表明,SCAE-ACGAN的故障诊断性能好,在样本数量少与工况变化情况下具有较好的健壮性和适应性。     

Microstructure and Mechanical Properties of Hot-Pressed Silicon Carbide-Aluminum Nitride Compositions

A flexural strength of up to 1 GPa was achieved in SiC-AIN materials and is attributed to a dense, equiaxial grain structure of the 2H(δ) SiC-AIN solid solution, with a relatively uniform grain size of ∼ 1 μm. The strength was found to decrease with increasing grain size. While the β→α phase transformation and the formation of various metastable polytypes make microstructural control difficult in SiC materials, excellent control is facilitated in SiC-AIN materials as a result of the stable 2H solid solution. Several mechanisms of grain refinement during the β→ 2H transition were observed, most notably the direct formation of several 2H grains from a single β grain. In addition, grain growth is limited by the diffusion-controlled nature of the transition. These mechanisms could be utilized to achieve even higher strength values, with potentially higher reliability of the materials in structural applications.     

Formation of β-SiAION as an Intermediate Oxidation Product of SiC–AIN Ceramics

The oxidation mechanisms in SiC-AlN solid solutions, exposed at 1370°C for 200 h in air, have been investigated by scanning electron microscopy (SEM), analytical and high-resolution electron microscopy (HREM), and X-ray diffraction techniques. Five hot-pressed compositions, including pure SiC and pure AIN, were examined. Assuming parabolic rate behavior, the values of the parabolic rate constants based on oxide layer thickness increased by over 2 orders of magnitude by increasing AIN contents from 0% to 62%. A continuous amorphous phase containing cristobalite and mullite crystallites was present at the oxide layer in 13% AIN and 27% AIN samples. Two reaction layers were found in the oxidized SiC–62% AIN solid solution. The inner layer consisted of a mixture of β-SiAION and graphite phases, while the outer layer consisted of mullite as a major phase and cristobalite as a minor phase. The oxidation proceeded in two stages: a partial oxidation of SiC–AIN to β-SiAION and carbon (graphite), and a final oxidation of β-SiAION to mullite + SiO₂.     

Static and Cyclic Fatigue of Alumina at High Temperatures: II, Failure Analysis

Failure mechanisms of an alumina, tested at 1200°C under static and various cyclic loading conditions, were examined. Slow crack growth of a single crack is the dominant mechanism for the failure in specimens under cyclic loading with a short duration of maximum stress at all applied stress levels, as well as at high applied loads for static loading and cyclic loading with a longer hold time at maximum stress. At low stress levels, failure of static loading and cyclic loading with a longer hold time at maximum stress might occur by formation and/or growth of multiple macrocracks. More importantly, for all the given loading conditions. The viscous glassy phase behind the crack tip could have a bridging effect on the crack surfaces. A simplified model for calculating effective stress intensity factor at the crack tip under static and various cyclic loading demonstrated a trend consistent with the stress–life data.     

Microstructure and Microchemistry of Polymer-Derived Crystalline SiC Fibers

Stoichiometric and near-stoichiometric silicon carbide fibers, derived from the polymer polycarbosilane, have been characterized by scanning Auger and transmission electron microscope techniques. Excess carbon usually appeared in the fibers, but it could be eliminated by controlling processing conditions. The fibers consisted mainly of submicrometer-sized SiC microcrystals, with a considerably twinned and stacking-faulted β-SiC structure. Free carbon, which appeared in the form of graphite, was the only detectable intergranular phase in the nonstoichioinetric SiC fibers, while the stoichiometric SiC fibers contained no detectable second phase whatsoever.     

可重构阵列自主容错方法研究

设计了一种具有故障自诊断和自修复能力的可重构阵列单元结构。在功能细胞单元内部实现分布式的故障自诊断,在测试到故障后,可以自主地将距故障单元最近的空闲单元进行替换,接着自动取消受故障影响的线网,并在新的布线端点间对这些线网重新布线。以4位并行乘法器为例,实验结果证明了可重构单元阵列的故障自修复能力,并验证其重构时间较短且可靠性较高。     

胚胎型仿生电路中具有自修复性能的存储器设计

胚胎型仿生电路是一种芯片级容错数字电路,其工作机制为汲取生物体生长过程的灵感从而使得数字系统具有自主容错能力。该电路是由二维电子细胞阵列组成,文中分析了这种数字电路的结构和容错机制,讨论了细胞的主要组成部件的设计,并基于海明码(Hamming Code)为细胞中的存储器设计了一种具有自检测和自修复能力的结构。最后,以4×4乘法器为应用实例,通过仿真实验验证了细胞中存储器的自诊断、自修复能力和该仿生电路的自修复能力。     

精密测量相机畸变测量及数据处理

本文介绍了一种精密测量相机畸变与主距的测量及数据处理方法,介绍了这种测量原理,仪器装置,分析了其测量精度,同时给出了一种通过用正交多项式对常用的最小二乘法拟合方法的改进来增加畸变曲线拟合精库的算法。本文所测量的对象是长春光机学院正在研制的“高空CCD立靶”测量系统。     

层次立方体(H-Cube)的自适应计算

多维数据立方（Cube）的计算对联机分析处理（online analytical processing，OLAP）有着极为重要的作用。针对传统的多维数据Cube计算中的不足，提出了自适应的计算方法，并给出了带有层次的Cube计算算法HCA算法。     

电池组均衡技术研究现状

电池组均衡技术是电池管理系统关键技术之一,可根据电池组状态信息对电池组进行实时均衡管理,避免不一致性的扩大,提高电池组容量利用率,能在一定程度上减缓电池组老化速度。本文深入分析了国内外的电池组均衡方法,按照均衡电路、控制策略以及系统结构3个方面,比较了各种均衡方法的优缺点及其适用场合,列举了不同方法的研究情况,给出了串联电池组均衡技术研究亟待解决的问题与发展方向展望。