The outlet flow fields of a low-speed repeating-stage compressor with bowed stator stages are measured with five-hole probe under the near stall condition when the rotor/stator axial gap varies. The performances of th...The outlet flow fields of a low-speed repeating-stage compressor with bowed stator stages are measured with five-hole probe under the near stall condition when the rotor/stator axial gap varies. The performances of the straight stator stages are investigated and compared to those of the bowed stator stages. The results show that using bowed stator stages could alleviate the flow separation at both upper and low corners of the suction surface and the endwalls, and decrease the losses along the flow passage as well as the outlet flow angle. As the rotor/stator axial gap decreases, although the diffusion capacity of the compressor increases obviously, the outlet flow field in the straight stator stages deteriorates quickly. By contrast, little changes occur in the bowed stator stages, indicating that as the rotor/stator axial gap decreases, improved performance is achieved in the bowed stator stages.展开更多
针对一般聚类获得的码本缺乏判别性表示导致不能有效进行人体动作识别的问题,提出了一种新的自适应码本学习方法,该方法将判别式词袋(bag of words,Bo W)动作表示和自适应码本学习结合,增强了码本的表示能力和特征的判别性。为了有效求...针对一般聚类获得的码本缺乏判别性表示导致不能有效进行人体动作识别的问题,提出了一种新的自适应码本学习方法,该方法将判别式词袋(bag of words,Bo W)动作表示和自适应码本学习结合,增强了码本的表示能力和特征的判别性。为了有效求解非凸目标函数,提出基于轮换优化迭代方法,即固定码本更新判别矩阵,然后判别矩阵更新固定码本,直至满足终止迭代条件,该方法为自适应码本学习提供了技术支持。仿真实验采用KTH、Hollywood2、芭蕾、i3Dpost数据库进行判别比较,识别率比现有典型方法平均提高了4%左右,学习到的码本在特征空间中具有良好的判别性能。相比于基于光流、方向梯度直方图(histograms of oriented gradients,HOG)等方法,计算复杂度更低,实用性更好。展开更多
基金National Natural Science Foundation of China (50646021)Chinese Specialized Research Fund for the Doctoral Pro-gram of Higher Education (20060213007)
文摘The outlet flow fields of a low-speed repeating-stage compressor with bowed stator stages are measured with five-hole probe under the near stall condition when the rotor/stator axial gap varies. The performances of the straight stator stages are investigated and compared to those of the bowed stator stages. The results show that using bowed stator stages could alleviate the flow separation at both upper and low corners of the suction surface and the endwalls, and decrease the losses along the flow passage as well as the outlet flow angle. As the rotor/stator axial gap decreases, although the diffusion capacity of the compressor increases obviously, the outlet flow field in the straight stator stages deteriorates quickly. By contrast, little changes occur in the bowed stator stages, indicating that as the rotor/stator axial gap decreases, improved performance is achieved in the bowed stator stages.
文摘针对一般聚类获得的码本缺乏判别性表示导致不能有效进行人体动作识别的问题,提出了一种新的自适应码本学习方法,该方法将判别式词袋(bag of words,Bo W)动作表示和自适应码本学习结合,增强了码本的表示能力和特征的判别性。为了有效求解非凸目标函数,提出基于轮换优化迭代方法,即固定码本更新判别矩阵,然后判别矩阵更新固定码本,直至满足终止迭代条件,该方法为自适应码本学习提供了技术支持。仿真实验采用KTH、Hollywood2、芭蕾、i3Dpost数据库进行判别比较,识别率比现有典型方法平均提高了4%左右,学习到的码本在特征空间中具有良好的判别性能。相比于基于光流、方向梯度直方图(histograms of oriented gradients,HOG)等方法,计算复杂度更低,实用性更好。
