Thermal analysis of beamline heat load components due to increased power delivery from EAST neutral beam injector

Journal of Thermal Analysis and Calorimetry - According to EAST’s experimental plan, the long-pulse and high-power operation of neutral beam injector is the requirement for EAST. In this...     

基于神经网络的二维不变性目标识别方法

提出了一种基于多层前馈神经网络的二维不变性目标识别方法。利用傅里叶描述器提取具有旋转、平移及尺度不变性的目标形状特征。由于所识别的工业工具具有一个自由度,它们形状有一定的动态变化导致同一目标的形状特征敌意一的不唯一性。文中采用含有两个隐层的多层前馈网络学习及识别这些特征矢量。在实验中,在四类机械工具进行测试,并将所提出方法与最近邻分类器进行比较,结果表明,具有反向传输（ＢＰ）学习算法的多层前馈网络     

从二维视图识别三维目标的多网络融合方法

提出了一种从二维视图识别三维目标的多网络融合方法,基于单个网络分类的置信度概念,有效地结合多个网络的输出结果作出最终分类判决,应用三个多层前向网络（隐层神经元数,初始权值等取不同值）,设计了基于分类确认度的多网络融合结构,对四类车辆目标进行的识别实验表明,所提出的多网络融合方法明显优于单个网络的识别性能。     

神经网络自动目标识别技术述评(二)

论述几种生物激励的神经网络模型，讨论它们在ＡＴＲ问题方面所取得的进展，最后给出本文结论。     

层析三维数字化仪层析图像滤波器的设计及研究

提出一种新型的递归中值滤波器，抛掉了统计参数的制约，将滤波算法转化为一种优化处理。该方法兼顾了滤波处理的光滑连续性及抑制噪声的累积特性，可有效地消除脉冲型干扰的影响，同时也从理论的角度上对该算法进行了分析。为消除加性高斯噪声，提出了一种基于图像边缘方向的小波线性滤波器，它仅仅处理边缘信息。该算汉的极大优点是克服了边缘模糊效应，小波的去噪逆向重构的处理方法对边缘为跃型的层析图象非常实用。     

Compensation of errors of alignment and contact pattern repositioning in hypoid gears with low crossing shaft angle

The investigation of the influence of different errors of alignment on the mesh behavior of hypoid gears with low crossing shaft angle has been carried out. As a result, a simplified methodology of compensation of errors of alignment and correction of the contact pattern by controlled axial displacements to the pinion and wheel of the hypoid gear set has been proposed. The shaft angle error and the offset error mainly influence the position of the contact pattern in profile direction. The pinion and wheel axial position errors influence the position of contact pattern along the tooth trace direction and change the position slightly along the tooth profile direction. All the assembly errors increase the peak-to-peak value of transmission errors, except positive values of the pinion axial position error that allows decreasing the peak-to-peak transmission error slightly in an appropriate range. Finite element analysis has been performed to verify the compensating ability of the developed methodology.