In this paper, a learning and recognition approach is proposed for univariate time series composed of output measurements of general nonlinear dynamical systems. Firstly, a class of dynamical systems in the canonical ...In this paper, a learning and recognition approach is proposed for univariate time series composed of output measurements of general nonlinear dynamical systems. Firstly, a class of dynamical systems in the canonical form is derived to describe the univariate time series by introducing coordinate transformation. An observer-based deterministic learning technique is then adopted to achieve dynamical modeling of the associated transformed systems of the training univariate time series, and the modeling results in the form of radial basis function network (RBFN) models are stored in a pattern library. Subsequently, multiple observer-based dynamical estimators containing the RBFN models in the pattern library are constructed for a test univariate time series, and a recognition decision scheme is proposed by the derived recognition indicator. On this basis, more concise recognition conditions are provided, which is beneficial for verifying the recognition results. Finally, simulation studies on the Rossler system and aero-engine stall warning verify the effectiveness of the proposed approach.展开更多
奇异值特征向量是用于图像识别的有效代数特征,但直接用奇异值特征向量做匹配进行人脸识别,识别率极低。通过对人脸图像奇异值向量和其对应的左右正交特征矩阵分析,发现图像的奇异值向量与图像的灰度范围具有相关性,即最大奇异值反映了...奇异值特征向量是用于图像识别的有效代数特征,但直接用奇异值特征向量做匹配进行人脸识别,识别率极低。通过对人脸图像奇异值向量和其对应的左右正交特征矩阵分析,发现图像的奇异值向量与图像的灰度范围具有相关性,即最大奇异值反映了图像灰度范围的位置,其他奇异值反映了灰度范围的宽度,而且与图像奇异值向量对应的左右正交特征矩阵能够表现图像轮廓的结构信息。基此,提出基于奇异值分解(singular value distribution,SVD)的基空间人脸识别算法,并通过ORL和ORL-IC数据库进行仿真,实验结果分析证明了图像的左右正交特征矩阵能够表现图像轮廓的结构信息。展开更多
Aiming at the low recognition accuracy of non-negative matrix factorization(NMF)in practical application,an improved spare graph NMF(New-SGNMF)is proposed in this paper.New-SGNMF makes full use of the inherent geometr...Aiming at the low recognition accuracy of non-negative matrix factorization(NMF)in practical application,an improved spare graph NMF(New-SGNMF)is proposed in this paper.New-SGNMF makes full use of the inherent geometric structure of image data to optimize the basis matrix in two steps.A threshold value s was first set to judge the threshold value of the decomposed base matrix to filter the redundant information in the data.Using L2 norm,sparse constraints were then implemented on the basis matrix,and integrated into the objective function to obtain the objective function of New-SGNMF.In addition,the derivation process of the algorithm and the convergence analysis of the algorithm were given.The experimental results on COIL20,PIE-pose09 and YaleB database show that compared with K-means,PCA,NMF and other algorithms,the proposed algorithm has higher accuracy and normalized mutual information.展开更多
In this paper, an improved radial basis function networks named hidden neuron modifiable radial basis function (HNMRBF) networks is proposed for target classification, and evolutionary programming (EP) is used as a le...In this paper, an improved radial basis function networks named hidden neuron modifiable radial basis function (HNMRBF) networks is proposed for target classification, and evolutionary programming (EP) is used as a learning algorithm to determine and modify the hidden neuron of HNMRBF nets. The result of passive sonar target classification shows that HNMRBF nets can effectively solve the problem of traditional neural networks, i. e. learning new target patterns on line will cause forgetting of the old patterns.展开更多
For acquiring the flow regime information of two-phase flow,a flow regime identification method using the Hilbert-Huang Transform (HHT) combined with Radial Basis Function neural networks was put forward.In this study...For acquiring the flow regime information of two-phase flow,a flow regime identification method using the Hilbert-Huang Transform (HHT) combined with Radial Basis Function neural networks was put forward.In this study,oil-gas two-phase flow in horizontal pipe was taken as the experimental object, differential pressure signals coming from Venturi tube were handled by Hilbert-Huang Transform,and characteristic vector closely associated with the flow regime were obtained.Flow regime was identified by using Radial Basis Function neural networks.While oil flux was in the range of 4.2 to 7.0 m3·h -1 and gas flux was 0 to 30 m3·h -1, this method showed high identification precision for bubble flow, slug flow, churn flow and annular flow et al.The experimental study showed that this method could precisely identify the flow regime and could be used easily.展开更多
基金supported by the National Postdoctoral Researcher Program of China(No.GZC20231451)the National Natural Science Foundation of China(Nos.61890922,62203263)the Shandong Province Natural Science Foundation(Nos.ZR2020ZD40,ZR2022QF062).
文摘In this paper, a learning and recognition approach is proposed for univariate time series composed of output measurements of general nonlinear dynamical systems. Firstly, a class of dynamical systems in the canonical form is derived to describe the univariate time series by introducing coordinate transformation. An observer-based deterministic learning technique is then adopted to achieve dynamical modeling of the associated transformed systems of the training univariate time series, and the modeling results in the form of radial basis function network (RBFN) models are stored in a pattern library. Subsequently, multiple observer-based dynamical estimators containing the RBFN models in the pattern library are constructed for a test univariate time series, and a recognition decision scheme is proposed by the derived recognition indicator. On this basis, more concise recognition conditions are provided, which is beneficial for verifying the recognition results. Finally, simulation studies on the Rossler system and aero-engine stall warning verify the effectiveness of the proposed approach.
文摘奇异值特征向量是用于图像识别的有效代数特征,但直接用奇异值特征向量做匹配进行人脸识别,识别率极低。通过对人脸图像奇异值向量和其对应的左右正交特征矩阵分析,发现图像的奇异值向量与图像的灰度范围具有相关性,即最大奇异值反映了图像灰度范围的位置,其他奇异值反映了灰度范围的宽度,而且与图像奇异值向量对应的左右正交特征矩阵能够表现图像轮廓的结构信息。基此,提出基于奇异值分解(singular value distribution,SVD)的基空间人脸识别算法,并通过ORL和ORL-IC数据库进行仿真,实验结果分析证明了图像的左右正交特征矩阵能够表现图像轮廓的结构信息。
基金This work was supported by the National Natural Science Foundation of China(Grant No.61501005)the Anhui Natural Science Foundation(Grant No.1608085 MF 147)+2 种基金the Natural Science Foundation of Anhui Universities(Grant No.KJ2016A057)the Industry Collaborative Innovation Fund of Anhui Polytechnic University and Jiujiang District(Grant No.2021cyxtb4)the Science Research Project of Anhui Polytechnic University(Grant No.Xjky2020120).
文摘Aiming at the low recognition accuracy of non-negative matrix factorization(NMF)in practical application,an improved spare graph NMF(New-SGNMF)is proposed in this paper.New-SGNMF makes full use of the inherent geometric structure of image data to optimize the basis matrix in two steps.A threshold value s was first set to judge the threshold value of the decomposed base matrix to filter the redundant information in the data.Using L2 norm,sparse constraints were then implemented on the basis matrix,and integrated into the objective function to obtain the objective function of New-SGNMF.In addition,the derivation process of the algorithm and the convergence analysis of the algorithm were given.The experimental results on COIL20,PIE-pose09 and YaleB database show that compared with K-means,PCA,NMF and other algorithms,the proposed algorithm has higher accuracy and normalized mutual information.
文摘In this paper, an improved radial basis function networks named hidden neuron modifiable radial basis function (HNMRBF) networks is proposed for target classification, and evolutionary programming (EP) is used as a learning algorithm to determine and modify the hidden neuron of HNMRBF nets. The result of passive sonar target classification shows that HNMRBF nets can effectively solve the problem of traditional neural networks, i. e. learning new target patterns on line will cause forgetting of the old patterns.
文摘For acquiring the flow regime information of two-phase flow,a flow regime identification method using the Hilbert-Huang Transform (HHT) combined with Radial Basis Function neural networks was put forward.In this study,oil-gas two-phase flow in horizontal pipe was taken as the experimental object, differential pressure signals coming from Venturi tube were handled by Hilbert-Huang Transform,and characteristic vector closely associated with the flow regime were obtained.Flow regime was identified by using Radial Basis Function neural networks.While oil flux was in the range of 4.2 to 7.0 m3·h -1 and gas flux was 0 to 30 m3·h -1, this method showed high identification precision for bubble flow, slug flow, churn flow and annular flow et al.The experimental study showed that this method could precisely identify the flow regime and could be used easily.
