The smart distribution system is the critical part of the smart grid, which also plays an important role in the safe and reliable operation of the power grid. The self-healing function of smart distribution network wi...The smart distribution system is the critical part of the smart grid, which also plays an important role in the safe and reliable operation of the power grid. The self-healing function of smart distribution network will effectively improve the security, reliability and efficiency, reduce the system losses, and promote the development of sustainable energy of the power grid. The risk identification process is the most fundamental and crucial part of risk analysis in the smart distribution network. The risk control strategies will carry out on fully recognizing and understanding of the risk events and the causes. On condition that the risk incidents and their reason are identified, the corresponding qualitative / quantitative risk assessment will be performed based on the influences and ultimately to develop effective control measures. This paper presents the concept and methodology on the risk identification by means of Hidden Semi-Markov Model (HSMM) based on the research of the relationship between the operating characteristics/indexes and the risk state, which provides the theoretical and practical support for the risk assessment and risk control technology.展开更多
As a kind of statistical method, the technique of Hidden Markov Model （HMM） is widely used for speech recognition. In order to train the HMM to be more effective with much less amount of data, the Subspace Distribut...As a kind of statistical method, the technique of Hidden Markov Model （HMM） is widely used for speech recognition. In order to train the HMM to be more effective with much less amount of data, the Subspace Distribution Clustering Hidden Markov Model （SDCHMM）, derived from the Continuous Density Hidden Markov Model （CDHMM）, is introduced. With parameter tying, a new method to train SDCHMMs is described. Compared with the conventional training method, an SDCHMM recognizer trained by means of the new method achieves higher accuracy and speed. Experiment results show that the SDCHMM recognizer outperforms the CDHMM recognizer on speech recognition of Chinese digits.展开更多
为了进行连续马尔可夫模型的初值提取,提出一种各类在训练样本空间近似均衡分布的K均值聚类法。在聚类的过程中引入惩罚因子,从而限制过多的训练矢量集中于一个或几个类,使样本空间划分近似均匀。连续马尔可夫模型初值提取实验证明,该...为了进行连续马尔可夫模型的初值提取,提出一种各类在训练样本空间近似均衡分布的K均值聚类法。在聚类的过程中引入惩罚因子,从而限制过多的训练矢量集中于一个或几个类,使样本空间划分近似均匀。连续马尔可夫模型初值提取实验证明,该方法与标准的K均值聚类法、LBG(L inde Buzo G ray)聚类法相比,降低了矢量量化产生的全局失真,各个类在样本空间的分布更加均匀,提高了矢量量化的性能。将该方法用于孤立词识别连续马尔可夫模型的初值提取,可使各个高斯概率密度函数的参数估计更逼近其无偏估计,从而提高了马尔可夫模型初值的可靠性。展开更多
Shearlet变换作为后小波时代的一个重要的多尺度几何分析工具具有良好的各向异性和方向捕捉性,同时它也可以对诸如图像等多维信号进行一种近最优的稀疏表示.非下采样Shearlet变换(NSST)在保持Shearlet变换特性的同时还具有平移不变特性...Shearlet变换作为后小波时代的一个重要的多尺度几何分析工具具有良好的各向异性和方向捕捉性,同时它也可以对诸如图像等多维信号进行一种近最优的稀疏表示.非下采样Shearlet变换(NSST)在保持Shearlet变换特性的同时还具有平移不变特性,这在具有丰富纹理和细节信息的图像处理中发挥着重要作用.该文首先对图像NSST方向子带内系数的概率密度分布进行分析,获得系数的稀疏统计特性和Cauchy分布拟合子带内系数的有效性;其次对NSST方向子带间系数的联合概率分布进行分析,获得方向子带系数间所具有的持续和传递特性,确定了一种NSST子带间树形架构的系数对应关系,进而提出一种NSST域隐马尔可夫模树模型(C-NSSTHMT),该模型通过Cauchy分布来拟合NSST系数,更好地揭示图像NSST变换后相同尺度子带内和不同尺度子带间系数的相关性.进一步提出一种基于所提出C-NSST-HMT模型的图像去噪算法,该算法对于含噪声方差为30和40的噪声图像,其去噪后的PSNR(Peak Signal to Noise Ratio)较NSCT-HMT方法分别提高了1.995dB和1.193dB.特别对纹理和细节丰富的图像,该算法在去噪的同时,有效地保留了图像的几何信息.展开更多
文摘The smart distribution system is the critical part of the smart grid, which also plays an important role in the safe and reliable operation of the power grid. The self-healing function of smart distribution network will effectively improve the security, reliability and efficiency, reduce the system losses, and promote the development of sustainable energy of the power grid. The risk identification process is the most fundamental and crucial part of risk analysis in the smart distribution network. The risk control strategies will carry out on fully recognizing and understanding of the risk events and the causes. On condition that the risk incidents and their reason are identified, the corresponding qualitative / quantitative risk assessment will be performed based on the influences and ultimately to develop effective control measures. This paper presents the concept and methodology on the risk identification by means of Hidden Semi-Markov Model (HSMM) based on the research of the relationship between the operating characteristics/indexes and the risk state, which provides the theoretical and practical support for the risk assessment and risk control technology.
基金Supported by the National Natural Science Foundation of China (No.60172048)
文摘As a kind of statistical method, the technique of Hidden Markov Model （HMM） is widely used for speech recognition. In order to train the HMM to be more effective with much less amount of data, the Subspace Distribution Clustering Hidden Markov Model （SDCHMM）, derived from the Continuous Density Hidden Markov Model （CDHMM）, is introduced. With parameter tying, a new method to train SDCHMMs is described. Compared with the conventional training method, an SDCHMM recognizer trained by means of the new method achieves higher accuracy and speed. Experiment results show that the SDCHMM recognizer outperforms the CDHMM recognizer on speech recognition of Chinese digits.
文摘为了进行连续马尔可夫模型的初值提取,提出一种各类在训练样本空间近似均衡分布的K均值聚类法。在聚类的过程中引入惩罚因子,从而限制过多的训练矢量集中于一个或几个类,使样本空间划分近似均匀。连续马尔可夫模型初值提取实验证明,该方法与标准的K均值聚类法、LBG(L inde Buzo G ray)聚类法相比,降低了矢量量化产生的全局失真,各个类在样本空间的分布更加均匀,提高了矢量量化的性能。将该方法用于孤立词识别连续马尔可夫模型的初值提取,可使各个高斯概率密度函数的参数估计更逼近其无偏估计,从而提高了马尔可夫模型初值的可靠性。
文摘Shearlet变换作为后小波时代的一个重要的多尺度几何分析工具具有良好的各向异性和方向捕捉性,同时它也可以对诸如图像等多维信号进行一种近最优的稀疏表示.非下采样Shearlet变换(NSST)在保持Shearlet变换特性的同时还具有平移不变特性,这在具有丰富纹理和细节信息的图像处理中发挥着重要作用.该文首先对图像NSST方向子带内系数的概率密度分布进行分析,获得系数的稀疏统计特性和Cauchy分布拟合子带内系数的有效性;其次对NSST方向子带间系数的联合概率分布进行分析,获得方向子带系数间所具有的持续和传递特性,确定了一种NSST子带间树形架构的系数对应关系,进而提出一种NSST域隐马尔可夫模树模型(C-NSSTHMT),该模型通过Cauchy分布来拟合NSST系数,更好地揭示图像NSST变换后相同尺度子带内和不同尺度子带间系数的相关性.进一步提出一种基于所提出C-NSST-HMT模型的图像去噪算法,该算法对于含噪声方差为30和40的噪声图像,其去噪后的PSNR(Peak Signal to Noise Ratio)较NSCT-HMT方法分别提高了1.995dB和1.193dB.特别对纹理和细节丰富的图像,该算法在去噪的同时,有效地保留了图像的几何信息.
