半监督多图嵌入的高光谱影像特征提取

针对传统图嵌入方法仅采用单一图结构无法有效表征高维数据中复杂本征结构,本文提出了一种半监督多图嵌入(SSMGE)方法,并应用于高光谱影像特征提取。该方法首先利用标记样本的类内、类间近邻点来构建类内超图、类间超图、类内普通图、类间普通图,然后通过无标记样本的近邻点和远离点构建无监督本征超图和惩罚超图,并以多图协同方式来表征高维数据间的复杂几何关系,实现鉴别特征提取。本文提出的SSMGE方法不仅能有效揭示数据点间超图和普通图的结构,而且在低维嵌入空间中增强同类数据聚集性和非同类数据的远离性,提取的鉴别特征可改善地物分类精度。在PaviaU和Urban高光谱数据集上进行了分类实验,本文方法的总体分类精度分别可达到85.92%和79.74%。相比普通图嵌入和超图方法,该算法明显提升了地物的分类性能。     

面向高光谱影像分类的多特征流形鉴别嵌入

鉴于传统维数约减方法对高光谱遥感影像进行降维时,往往只利用了单一的光谱特征,限制了分类性能的提升。提出一种基于多特征流形鉴别嵌入的维数约减方法,该方法首先提取高光谱数据的LBP(Local Binary Patterns)纹理特征,然后利用样本点的光谱-LBP特征联合距离及类别信息构建类内图和类间图以发现高光谱影像中的鉴别流形结构,在低维嵌入空间中不仅保持来自同一像素的光谱和纹理特征的相似性,而且使同类点尽可能紧致、不同类点远离,实现空-谱联合低维鉴别特征提取,以有效提高地物分类性能。在Indian Pines和黑河高光谱遥感数据集上的实验表明,本文算法的分类精度在不同实验条件下均优于传统的维数约减方法,其分类精度可达95.05%和96.20%,在较少训练样本条件下优势更为明显,有利于实际应用。     

面向高光谱影像分类的深度流形重构置信网络

鉴于传统深度学习方法只提取了高光谱图像中的深度抽象信息,而未能充分揭示样本之间的局部几何结构关系,限制了分类性能的提升,本文提出了一种新的特征提取网络——深度流形重构置信网络。该网络首先通过深度置信网络提取深度抽象特征,为进一步增强抽象特征的鉴别能力,在图嵌入框架下通过样本数据的邻域点和各邻域的同类近邻重构点来构建类内图和类间图,并在低维空间中分离类间近邻点与其重构点的同时压缩类内近邻点和相应的重构点,实现提取深度鉴别特征,以改善不同类数据的可分性,进而提升地物分类精度。在KSC和MUUFL Gulfport高光谱数据集上的实验结果表明,本文算法的总体分类精度分别达到了94.71%和86.38%。相比较其他算法,本文算法有效提升了地物分类能力,更有利于实际应用。     

应用相关近邻局部线性嵌入算法的高光谱遥感影像分类

传统的局部线性嵌入（LLE）算法需用欧氏距离度量近邻,但欧氏距离只表示两点间的直线距离,在高维空间中不一定能反映数据间的真实空间分布,导致近邻选取不稳定。针对此问题,本文提出了相关近邻（CN）LLE（CN-LLE）和相关最近邻分类（CNN）算法。提出的算法首先利用相关系数度量数据间的近邻,实现更准确的局部重构,提取鉴别特征;然后用CNN对低维嵌入特征进行分类。在KSC和Indian Pine高光谱遥感数据集上的地物分类实验结果表明：本文提出的CN-LLE+CNN算法比LLE、LLE+CNN和CN-LLE等算法的总分类精度提升了2.11%~11.55%,Kappa系数提升了0.026~0.143。由于该算法增加了近邻为同类的概率,便于更有效地提取同类数据的鉴别特征,且有更好的稳定性,故能更有效地实现高光谱遥感数据的地物分类。     

多核融合多尺度特征的高光谱影像地物分类

对于高光谱影像地物分类问题,为更加有效地利用像元空间信息和光谱信息,提高地物分类精度,提出了多核融合多尺度特征的分类方法。首先,通过多尺度空间滤波和PCA白化,提取出多尺度特征;接着在核稀疏表示分类器内使用多核方式对分别表示每项特征,在分类器内实现特征自动融合,根据子核与理想核、子核之间距离求取核组合的权重,使用训练集所构成的字典在特征空间内对待测样本进行线性表示,根据每类地物的重构误差确定待测像元所属地物类别。实验结果表明:对于Indian Pines影像和Pavia University影像总体分类精度分别达到99.51%和97.96%,较传统方法明显提高,并且对于小样本地物识别精度也都能达到90%以上。本文算法对于高光谱影像地物具有更强的识别能力,并且具有较强的稳定性和鲁棒性。     

加权空-谱与最近邻分类器相结合的高光谱图像分类

提出了一种基于加权空-谱距离(WSSD)的相似性度量方法 ,并将其应用到最近邻分类器(KNN)中,导出了一种新的高光谱图像分类算法。该算法利用高光谱图像的物理特性,通过引入空间窗口和光谱因子这两个参数来挖掘出图像中的空间信息与光谱信息,利用空间近邻点对中心像元进行重构。在最大限度减少图像冗余信息的基础上,增大了同类像元间的相似性以及异类像元间的差异性,获得了更为有效的鉴别特征,从而更好地实现了数据间的相似性度量。基于Indian Pines和PaviaU高光谱数据集进行了实验,结果表明:将提出的WSSD-KNN算法应用于高光谱图像分类时,其分类精度高于其他算法,总体分类精度分别达到了91.72%和96.56%。由于算法较好地融合了图像中的空间-光谱信息,提取出了更为有效的鉴别特征,故不仅有效地改善了高光谱数据的地物分类精度,而且可在训练样本较少时,保持较高的识别率。     

基于半监督稀疏鉴别嵌入的高光谱遥感影像分类

为了有效利用已标记与未标记样本提高高光谱遥感影像分类精度,提出一种新的半监督流形学习方法——半监督稀疏鉴别嵌入算法(SSDE)。该算法结合了近邻流形结构及稀疏性的优点,不仅保留样本间的稀疏重构关系,而且通过引入少量有标记的训练样本以及大量无标记训练样本来获得高维数据的内在属性以及低维流形结构,实现鉴别特征提取,提高分类精度。在Washington DC Mall和Indian Pine数据集上的分类识别实验表明,该算法能够较为有效地发现高维空间中数据的内蕴结构,分类性能比其他算法有明显的提升。在随机选取8个有类别标记和60个无类别标记的数据作为训练样本的情况下,本文提出的SSDE算法在上述两个数据集上的分类精度分别达到了77.36%和97.85%。     

基于多尺度分割的高光谱图像稀疏表示与分类

针对高光谱特征的稀疏表示,提出了一种基于多尺度分割的空间加权算法用于高光谱图像分类。该算法采用更合理的邻域定义挖掘空间先验信息,优化类边缘像元的稀疏表示。首先,通过多尺度分割提供邻域空间约束;结合拉普拉斯尺度混合(LSM)先验,分别对每个邻域组内像元进行空间加权的稀疏表示。然后,采用概率支持向量机(SVM)分类,同时提供像元的分类标签及其置信度。最后,以此置信度为权重,对多尺度分类图进行加权融合,生成最终的分类图。实验显示,本文算法能够增强光谱特征表示的稀疏性和鲁棒性,提高总体分类精度;在小样本训练下,单类的分类精度可提升30%左右,表明该算法在高光谱应用中具有较强的实用性。     

基于分数阶微分的高光谱图像特征提取与分类

针对高光谱遥感图像的特征提取与地物分类,提出一种基于分数阶微分的高光谱图像特征提取方法,设计二维分数阶微分掩模提取高光谱图像的像素空间分数阶微分（SpaFD）特征,并提出一种空谱联合准则用于选取微分掩模阶数。为充分利用高光谱图像的空间特征与光谱特征,将SpaFD特征与原始特征直连融合获得SpaFD-Spe-Spa混合特征,并采用三维卷积神经网络（3DCNN）、先采用主成分分析（PCA）对像素光谱进行降维处理再送入三维卷积神经网络（3DCNNPCA）以及采用混合光谱网络（HybridSN）验证SpaFD-Spe-Spa混合特征的有效性。实验中分别采用3×3,5×5和7×7的分数阶微分掩模进行空间特征提取,4个真实高光谱图像的实验结果表明,所提取的SpaFD特征和SpaFDSpe-Spa特征可有效提升高光谱图像的地物分类精度,且SpaFD-Spe-Spa特征对地物分类准确率的提升更为明显：SpaFD特征相比原始特征在Indian Pines,Botswana,Pavia University和Salinas 4个数据上的分类识别率在最优情况下分别提升了3.87%,1.42%,2.41%和2...     

面向高光谱图像分类的空谱判别分析

针对传统的基于特征提取的高光谱图像地物分类算法大多只考虑光谱信息而忽略空间信息的问题,提出了一种面向高光谱分类的半监督空谱全局与局部判别分析(S3 GLDA)算法。该算法首先利用少量标记样本保存数据集的线性可分性和全局判别信息,再依靠较多的无标记的空间局部近邻像元来揭示局部判别信息和非线性局部流形,使高光谱遥感图像的光谱域全局判别结构和空间域局部判别结构在低维特征空间同时得以保留,并在输出特征中自动融入了空间信息,构成了半监督的空谱判别分析。在Indian Pines和PaviaU数据集的实验表明,总体分类精度分别达到76.24%和82.96%。与现有几种算法比较,该算法有效提高了输出特征在低维空间的判别能力,更好地揭示了数据集的内在非线性多模本质,有效提升了高光谱图像数据集的地物分类精度。     

三绕组变压器的双绕组等效建模及在Matpower中的应用

针对实际电网存在着大量的三绕组变压器,但国际上一些著名商业软件,如BPA仿真软件、Matpower这一权威潮流计算开源软件,均只提供双绕组变压器模型,限制了其在具有三绕组变压器的电力系统中的应用的问题,潮流计算是自主开发的电力系统各种应用软件的核心模块,因此依托国际权威开源程序进行二次开发,是一种较好的选择。对Matpower要求的数据格式进行了归纳,对变压器的一般的等值电路及带理想变压器的等值电路和带标幺值的等值电路进行了分析研究,提出了三绕组变压器转换为双绕组等效模型的建模方法,使得原先只适应双绕组变压器的潮流计算软件可以适用于三绕组变压器电网的潮流计算;最后以Matpower软件为例进行了案例计算,并用PSASP仿真软件进行对比验证。研究结果证明,所提出的建模方法是有效的。     

On the Relation of Load to Average Gap in the Contact Between Surfaces with Longitudinal Roughness

A computer simulation model for the contact between longitudinally-oriented rough surfaces has been formulated. This model closely duplicates the actual surf ace contact deformation behavior by taking into account the elastic interactions between the asperities. There were no assumptions made about the shapes, or any deformation behavior of the asperities, except for their obeying the laws of elasticity. The plastic deformations on the high asperity peaks were taken into account by setting a ceiling on their contact pressures at the material hardness value. The simulations used real surface profiles which were digitized from unworn circumferentially ground steel surfaces. Each pair of these profiles was mathematically combined to form an equivalent rough profile pressing against an infinitely rigid flat and having the appropriately adjusted elastic modulus. A total of 28 different pairs of profiles were used in the simulations. Each contacting pair was subjected to 30 different load levels and the local contact pressures and deformations were calculated. The contact simulations yielded some important mathematical relationships between parameters, such as the real area of contact, average gap, and average asperity load through statistical curve fitting. Two analytical functions were generated to relate the average load to average gap and the real area of contact to load.     

电站锅炉模糊风险评价方法研究与应用

针对电站锅炉风险等级评价问题,将模糊综合评价技术应用到电站锅炉风险等级评价中。开展了电站锅炉失效影响因素及模糊评价因素的重要程度分析,建立了一套科学合理适于在线评价的电站锅炉风险评价体系,提出了模糊综合评价技术与改进的模糊层次分析法相结合的模糊风险评价方法,利用改进的模糊层次分析法计算了指标体系中各层指标权重。对某一台电站锅炉的实际运行工况影响子因素进行了模糊风险评价,采用模糊合成算子进行模糊综合运算得到电了站锅炉运行工况影响子因素的评价向量。研究结果表明:电站锅炉的运行工况影响子因素的风险评价等级为第6级,失效可能性等级为小,该电站锅炉运行工况良好。     

交通荷载作用下桥梁结构参数识别方法

应用动力节点加载法,将作用在桥梁上的交通荷载转化为等效节点荷载,再用两步构造法识别结构物理参数.第1步,用Newmark-β方法变换结构运动方程,并用结构的加速度响应求得变换空间内的位移、速度和加速度响应;第2步,基于最小二乘原理,构造出变换空间内求解结构参数的递推式,进行结构物理参数识别.数值模拟结果表明,在交通荷载作用下,该方法能够快速、准确地进行桥梁结构参数识别.     

RLS and LMS blind adaptive multi-user detection method and comparison in acoustic communication

RLS and LMS blind adaptive multi-user detection algorithm and multi-user detector was proposed to solve the problem of multi-user signal detection problem encountered in underwater acoustic communication networks.In simulation analysis,RLS and the LMS blind adaptive multi-user detector were designed and tested for synchronous and asynchronous multi-user communication process.The results of SIR comparison and MMSE comparison show that,both of the two methods can realize blind adaptive detection when any user change in multi-user communication,during this process,the training communication sequences are not needed.The RLS algorithm has about 5 dB higher in SIR compared with LMS algorithm,and the convergence velocity of RLS algorithm is also higher than LMS algorithm when the communication users change.RLS algorithm has better ability in multi-user detection than that of LMS algorithm,and it has great attraction and guiding significance for solving the problem of multiple access interference(MAI) in multi-user communication.     

Robust Impedance Control of Robots Using an Adaptive Interaction Force Observer

For robot interaction control,the interaction force between the robot and the manipulated object or environment should be monitored.Impedance control is a type ...     

Research on key techniques of expendable conductivity temperature depth measuring system

This paper analysis the developing of expendable conductivity temperature depth measuring system(XCTD)and introduce its principle of measuring about temperature,salinity and depth of ocean.Some key techniques are put forward.According to the real needs of XCTD,conductivity sensor with high sensitivity is designed by principle of electromagnetic induce,the ocean conductivity from induced electromotive force has been calculated.Adding temperature correction circuit would help to reduce error of conductivity measurement because of sharply changing temperature.Advanced temperature measuring circuit of high precision and the constant current source is used to weaken effect of self-heating of resistance and fluctuation of the source.On respect of remote data transmission,LVDS is a good choice for the purpose of guarantee the quality of data transmitted and the transmission distance is reaching to thousand meters in the seawater.Modular programming method is also brought into this research aimed at improve the stability,reliability and maintainability of the whole measuring system.In February,2015,the trials in South China Sea demonstrate that the developed XCTD realize effective measurement at a speed of 6 knots and detection depth at 800 m.The consistency coefficient of the acquired data is greater than 0.99 and the success rate of probe launching is above 90%.     

CHARACTERISTCS OF FLUID FILM IN OPTIMIZED SPIRAL GROOVE MECHANICAL SEAL

In order to investigate the sealing performance variation resulted from the thermal deformation of the end faces, the equations to calculate the fluid film pressure distribution, the bearing force and the leakage rate are derived, for the fluid film both in parallel gap and in wedgy gap. The geometrical parameters of the sealing members are optimized by means of heat transfer analysis and complex method. The analysis results indicate that the shallow spiral grooves can generate hydrodynamic pressure while the rotating ring rotates and the bearing force of the fluid film in spiral groove end faces is much larger than that in the flat end faces. The deformation increases the bearing force of the fluid film in flat end faces, but it decreases the hydrodynamic pressure of the fluid film in spiral groove end faces. The gap dimensions which determine the characteristics of the fluid film is obtained by coupling analysis of the frictional heat and the thermal deformation in consideration of the equilibrium condition of the bearing force and the closing force. For different gap dimensions, the relationship between the closing force and the leakage rate is also investigated, based on which the leakage rate can be controlled by adjusting the closing force.     

Quantum Cooperative Robotics and Autonomy

The intersection of Quantum Technologies and Robotics Autonomy is explored in the present paper.The two areas are brought together in establishing an interdisci...     

OPTIMIZATION METHOD ON IMPELLER MERIDIONAL CONTOUR AND 3D BLADE

An optimization method for 3D blade and meridional contour of centrifugal or mixed-flow impeller based on the 3D viscous computational fluid dynamics (CFD) analysis is proposed. The blade is indirectly parameterized using the angular momentum and calculated by inverse design method. The design variables are separated into two categories: the meridional contour design variables and the blade design variables. Firstly, only the blade is optimized using genetic algorithm with the meridional contour remained constant. The artificial neural network (ANN) techniques with the training sample data schemed according to design of experiment theory are adopted to construct the response relation between the blade design variables and the impeller performance. Then, based on the ANN approximated relation between the meridional contour design variables and impeller performance, the meridional contour is optimized. Fewer design variables and less calculation effort is required in this method that may be widely used in the optimization of three-dimension impellers. An optimized impeller in a mixed-flow pump, where the head and the efficiency are enhanced by 12.9% and 4.5% respectively, confirms the validity of this newly proposed method.