基于奇异值分解(singular value decomposition,SVD)和最小二乘支持向量机(least square support vector machine,LS-SVM)提出电能质量扰动类型识别的新方法。通过对电能质量扰动信号的小波包变换系数矩阵进行奇异值分解,将基频、扰动...基于奇异值分解(singular value decomposition,SVD)和最小二乘支持向量机(least square support vector machine,LS-SVM)提出电能质量扰动类型识别的新方法。通过对电能质量扰动信号的小波包变换系数矩阵进行奇异值分解,将基频、扰动频率分量、噪声分解到不同的正交特征子空间。再与正常电压信号的奇异值作比值以抵消噪声能量的影响,最大限度地体现出扰动类型间的细微差别,以此作为扰动特征向量,作为最小二乘支持向量机分类器的输入参数,来实现电能质量扰动类型的识别。仿真结果表明,该方法识别准确率高,受噪声影响小,算法稳定性好。展开更多
提出了一种基于连续小波变换(continuous walelet transform,CWT)和奇异值分解(singular value decomposition,SVD)相结合的提升小波系数SVD辨识信号振荡频率和模式信息提取及信号去噪的新方法。克服了噪声较大或者密集模态时,小波脊线...提出了一种基于连续小波变换(continuous walelet transform,CWT)和奇异值分解(singular value decomposition,SVD)相结合的提升小波系数SVD辨识信号振荡频率和模式信息提取及信号去噪的新方法。克服了噪声较大或者密集模态时,小波脊线不清晰甚至会出现混叠和交叉难以提取频率的情况,根据提升的小波系数奇异值分解频率向量识别各阶振荡模式的频率。同时选用小波能量系数来识别主导振荡模式,用小波软阈值去噪和SVD分解后矩阵重构来进行信号去噪。CWT可以处理含时变振荡模式的低频振荡信号,且对模式参数具有较高的辨识精度。仿真算例验证了算法的有效性和适用性。展开更多
将基于变量预测模型的模式识别(variable predictive model based class discriminate,简称VPMCD)方法、经验模态分解(empirical mode decomposition,简称EMD)方法和奇异值分解(singular value decomposition,简称SVD)相结合,提出了一...将基于变量预测模型的模式识别(variable predictive model based class discriminate,简称VPMCD)方法、经验模态分解(empirical mode decomposition,简称EMD)方法和奇异值分解(singular value decomposition,简称SVD)相结合,提出了一种基于EMD,SVD和VPMCD的齿轮故障的诊断方法。首先,对齿轮振动信号进行EMD分解,得到若干个IMF(intrinsic mode function,简称IMF)分量;其次,将包含齿轮主要故障信息的前几个IMF分量组成特征向量矩阵,并对其进行SVD分解;最后,将奇异值作为特征向量建立VPMCD多故障分类器,以此来区分齿轮的工作状态和故障类型。将提出的方法应用于齿轮实验数据,分析结果表明,该方法能够实现齿轮故障类型的分类和诊断,是一种有效可行的齿轮故障诊断方法。展开更多
Large reflector antennas are widely used in radars, satellite communication, radio astronomy, and so on. The rapid developments in these fields have created demands for development of better performance and higher sur...Large reflector antennas are widely used in radars, satellite communication, radio astronomy, and so on. The rapid developments in these fields have created demands for development of better performance and higher surface accuracy. However, low accuracy and low effi- ciency are the common disadvantages for traditional panel alignment and adjustment. In order to improve the surface accuracy of large reflector antenna, a new method is pre- sented to determinate panel adjustment values from far field pattern. Based on the method of Physical Optics （PO）, the effect of panel facet displacement on radiation field value is derived. Then the linear system is constructed between panel adjustment vector and far field pattern. Using the method of Singular Value Decomposition （SVD）, the adjustment value for all panel adjustors are obtained by solving the linear equations. An experiment is conducted on a 3.7 m reflector antenna with 12 segmented panels. The results of simulation and test are similar, which shows that the presented method is feasible. Moreover, thediscussion about validation shows that the method can be used for many cases of reflector shape. The proposed research provides the instruction to adjust surface panels efficiently and accurately.展开更多
Combustion instability of pilot flame has been investigated in a model pilot bluff body stabilized combustor by running the pilot flame only. The primary objectives are to investigate the pilot flame dynamics and to p...Combustion instability of pilot flame has been investigated in a model pilot bluff body stabilized combustor by running the pilot flame only. The primary objectives are to investigate the pilot flame dynamics and to provide bases for the study of the interaction mechanisms between the pilot flame and the main flame. Dynamic pressures are measured by dynamic pressure transduc- ers. A high speed camera with CH＊ bandpass filter is used to capture the pilot flame dynamics. The proper orthogonal decomposition （POD） is used to further analyze the high speed images. With the increase of the pilot fuel mass flow rate, the pilot flame changes from stable to unstable state grad- ually. The combustion instability frequency is 136 Hz when the pilot flame is unstable. Numerical simulation results show that the equivalence ratios in both the shear layer and the recirculation zone increase as the pilot fuel mass flow rate increases. The mechanism of the instability of the pilot flame can be attributed to the coupling between the second order acoustic mode and the unsteady heat release due to symmetric vortex shedding. These results illustrate that the pilot fuel mass flow rate has significant influences on the dynamic stability of the pilot flame.展开更多
Flame features and dynamics are important to the explanation and prediction of a lean blowout(LBO)phenomenon.In this paper,recognition of near-LBO flame features and oscillation characterization methods were proposed ...Flame features and dynamics are important to the explanation and prediction of a lean blowout(LBO)phenomenon.In this paper,recognition of near-LBO flame features and oscillation characterization methods were proposed based on flame spectroscopic images.High-speed planar laser-induced fluorescence measurements of OH were used to capture unique dynamic features such as the local extinction and reignition feature and entrained reactant pockets.The Zernike moment demonstrated a good performance in recognition of stability and near-LBO conditions,though the geometric moment had more advantages to characterize frequency characteristics.Low-frequency oscillations,especially at the obvious self-excited oscillation frequency around 200 Hz,were found when approaching an LBO condition,which can be expected to be used as a novel prediction characteristic parameter of the flameout limit.Proper orthogonal decomposition(POD)and dynamic mode decomposition(DMD)were used to conduct dynamic analysis of near-LBO flames.POD modes spectra showed the unique frequency characteristics of stable and near-LBO flames,which were basically in line with those at the heat-release frequency.The primary POD modes demonstrated that the radial vibration mode dominated in a stable flame,while the rotation mode was found to exist in a near-LBO flame.Analysis of modal decomposition showed that flame shedding and agminated entrained reactant pockets were responsible for generating self-excited flame oscillations.展开更多
针对动态对比度增强磁共振灌注成像中脑血容积的计算,提出基于Hankel矩阵的奇异值分解(Singular Value Decomposition,SVD)算法。在奇异值数目的确定上采用差分谱量级差的研究方法,对算法进行理论推导与仿真模拟,得到较为理想的滤波效...针对动态对比度增强磁共振灌注成像中脑血容积的计算,提出基于Hankel矩阵的奇异值分解(Singular Value Decomposition,SVD)算法。在奇异值数目的确定上采用差分谱量级差的研究方法,对算法进行理论推导与仿真模拟,得到较为理想的滤波效果。由于成像过程存在测量噪声的干扰,分析了信噪比和示踪剂延迟对算法的影响。仿真结果表明,信噪比越低(SNR=5 d B),算法处理效果越明显;信噪比增高(SNR=100 d B),估计值偏差减小,结果越为准确。且该算法不受示踪剂延迟的影响。与传统奇异值分解算法相比,采用基于Hankel矩阵的奇异值算法可以更为准确地估计脑血容积。展开更多
Rhesus monkey models of Parkinson＇s disease were induced by injection of N-methyl-4-phenyl- 1,2,3,6-tetrahydropyridine. Neural firings were recorded using microelectrodes placed in the interna segment of the globus p...Rhesus monkey models of Parkinson＇s disease were induced by injection of N-methyl-4-phenyl- 1,2,3,6-tetrahydropyridine. Neural firings were recorded using microelectrodes placed in the interna segment of the globus pallidus. The wavelets and power spectra show gradual power reduction during the disease process along with increased firing rates in the Parkinson＇s disease state. Singular values of coefficients decreased considerably during tremor-related activity as well as in the Parkinson＇s disease state compared with normal signals, revealing that higher-frequency components weaken when Parkinson＇s disease occurs. We speculate that the death of neurons could be reflected by irregular frequency spike trains, and that wavelet packet decomposition can effectively detect the degradation of neurons and the loss of information transmission in the neural circuitry.展开更多
文摘基于奇异值分解(singular value decomposition,SVD)和最小二乘支持向量机(least square support vector machine,LS-SVM)提出电能质量扰动类型识别的新方法。通过对电能质量扰动信号的小波包变换系数矩阵进行奇异值分解,将基频、扰动频率分量、噪声分解到不同的正交特征子空间。再与正常电压信号的奇异值作比值以抵消噪声能量的影响,最大限度地体现出扰动类型间的细微差别,以此作为扰动特征向量,作为最小二乘支持向量机分类器的输入参数,来实现电能质量扰动类型的识别。仿真结果表明,该方法识别准确率高,受噪声影响小,算法稳定性好。
文摘提出了一种基于连续小波变换(continuous walelet transform,CWT)和奇异值分解(singular value decomposition,SVD)相结合的提升小波系数SVD辨识信号振荡频率和模式信息提取及信号去噪的新方法。克服了噪声较大或者密集模态时,小波脊线不清晰甚至会出现混叠和交叉难以提取频率的情况,根据提升的小波系数奇异值分解频率向量识别各阶振荡模式的频率。同时选用小波能量系数来识别主导振荡模式,用小波软阈值去噪和SVD分解后矩阵重构来进行信号去噪。CWT可以处理含时变振荡模式的低频振荡信号,且对模式参数具有较高的辨识精度。仿真算例验证了算法的有效性和适用性。
文摘将基于变量预测模型的模式识别(variable predictive model based class discriminate,简称VPMCD)方法、经验模态分解(empirical mode decomposition,简称EMD)方法和奇异值分解(singular value decomposition,简称SVD)相结合,提出了一种基于EMD,SVD和VPMCD的齿轮故障的诊断方法。首先,对齿轮振动信号进行EMD分解,得到若干个IMF(intrinsic mode function,简称IMF)分量;其次,将包含齿轮主要故障信息的前几个IMF分量组成特征向量矩阵,并对其进行SVD分解;最后,将奇异值作为特征向量建立VPMCD多故障分类器,以此来区分齿轮的工作状态和故障类型。将提出的方法应用于齿轮实验数据,分析结果表明,该方法能够实现齿轮故障类型的分类和诊断,是一种有效可行的齿轮故障诊断方法。
基金Supported by National Natural Science Foundation of China(Grant Nos.51490661,51490660,51205301)National Key Basic Research Program of China(973 Program,Grant No.2015CB857100)Special Funding for Key Laboratory of Xinjiang Uygur Autonomous Region,China(Grant No.2014KL012)
文摘Large reflector antennas are widely used in radars, satellite communication, radio astronomy, and so on. The rapid developments in these fields have created demands for development of better performance and higher surface accuracy. However, low accuracy and low effi- ciency are the common disadvantages for traditional panel alignment and adjustment. In order to improve the surface accuracy of large reflector antenna, a new method is pre- sented to determinate panel adjustment values from far field pattern. Based on the method of Physical Optics （PO）, the effect of panel facet displacement on radiation field value is derived. Then the linear system is constructed between panel adjustment vector and far field pattern. Using the method of Singular Value Decomposition （SVD）, the adjustment value for all panel adjustors are obtained by solving the linear equations. An experiment is conducted on a 3.7 m reflector antenna with 12 segmented panels. The results of simulation and test are similar, which shows that the presented method is feasible. Moreover, thediscussion about validation shows that the method can be used for many cases of reflector shape. The proposed research provides the instruction to adjust surface panels efficiently and accurately.
文摘Combustion instability of pilot flame has been investigated in a model pilot bluff body stabilized combustor by running the pilot flame only. The primary objectives are to investigate the pilot flame dynamics and to provide bases for the study of the interaction mechanisms between the pilot flame and the main flame. Dynamic pressures are measured by dynamic pressure transduc- ers. A high speed camera with CH＊ bandpass filter is used to capture the pilot flame dynamics. The proper orthogonal decomposition （POD） is used to further analyze the high speed images. With the increase of the pilot fuel mass flow rate, the pilot flame changes from stable to unstable state grad- ually. The combustion instability frequency is 136 Hz when the pilot flame is unstable. Numerical simulation results show that the equivalence ratios in both the shear layer and the recirculation zone increase as the pilot fuel mass flow rate increases. The mechanism of the instability of the pilot flame can be attributed to the coupling between the second order acoustic mode and the unsteady heat release due to symmetric vortex shedding. These results illustrate that the pilot fuel mass flow rate has significant influences on the dynamic stability of the pilot flame.
基金supported by the Heilongjiang Provincial Natural Science Foundation of China(No.LH2021F028)。
文摘Flame features and dynamics are important to the explanation and prediction of a lean blowout(LBO)phenomenon.In this paper,recognition of near-LBO flame features and oscillation characterization methods were proposed based on flame spectroscopic images.High-speed planar laser-induced fluorescence measurements of OH were used to capture unique dynamic features such as the local extinction and reignition feature and entrained reactant pockets.The Zernike moment demonstrated a good performance in recognition of stability and near-LBO conditions,though the geometric moment had more advantages to characterize frequency characteristics.Low-frequency oscillations,especially at the obvious self-excited oscillation frequency around 200 Hz,were found when approaching an LBO condition,which can be expected to be used as a novel prediction characteristic parameter of the flameout limit.Proper orthogonal decomposition(POD)and dynamic mode decomposition(DMD)were used to conduct dynamic analysis of near-LBO flames.POD modes spectra showed the unique frequency characteristics of stable and near-LBO flames,which were basically in line with those at the heat-release frequency.The primary POD modes demonstrated that the radial vibration mode dominated in a stable flame,while the rotation mode was found to exist in a near-LBO flame.Analysis of modal decomposition showed that flame shedding and agminated entrained reactant pockets were responsible for generating self-excited flame oscillations.
文摘针对动态对比度增强磁共振灌注成像中脑血容积的计算,提出基于Hankel矩阵的奇异值分解(Singular Value Decomposition,SVD)算法。在奇异值数目的确定上采用差分谱量级差的研究方法,对算法进行理论推导与仿真模拟,得到较为理想的滤波效果。由于成像过程存在测量噪声的干扰,分析了信噪比和示踪剂延迟对算法的影响。仿真结果表明,信噪比越低(SNR=5 d B),算法处理效果越明显;信噪比增高(SNR=100 d B),估计值偏差减小,结果越为准确。且该算法不受示踪剂延迟的影响。与传统奇异值分解算法相比,采用基于Hankel矩阵的奇异值算法可以更为准确地估计脑血容积。
基金supported in part by a grant from the National Natural Science Foundation of China,No. 81071150,10872156the National High Technology Research and Development Program of China (863 Program),No.2006AA04Z370
文摘Rhesus monkey models of Parkinson＇s disease were induced by injection of N-methyl-4-phenyl- 1,2,3,6-tetrahydropyridine. Neural firings were recorded using microelectrodes placed in the interna segment of the globus pallidus. The wavelets and power spectra show gradual power reduction during the disease process along with increased firing rates in the Parkinson＇s disease state. Singular values of coefficients decreased considerably during tremor-related activity as well as in the Parkinson＇s disease state compared with normal signals, revealing that higher-frequency components weaken when Parkinson＇s disease occurs. We speculate that the death of neurons could be reflected by irregular frequency spike trains, and that wavelet packet decomposition can effectively detect the degradation of neurons and the loss of information transmission in the neural circuitry.
