二型Takagi-Sugeno-Kang模糊模型和不确定高斯混合模型的等价性

不确定的高斯混合模型和二型Takagi-Sugeno-Kang(TSK)模糊模型之间的对应关系被建立: 任何一个不确定的高斯混合模型都唯一对应着一个二型TSK模糊系统, 不确定的高斯混合模型的条件均值和二型TSK模糊 系统的输出是等价的. 基于此, 一种设计二型模糊系统的新方法被提出: 通过建立不确定的高斯混合模型确定二型TSK模糊系统, 即用概率统计的方法设计二型模糊系统. 仿真实验结果表明利用不确定高斯混合模型设计的二型模糊系统比其它模型具有更强的抗噪性和更快的速度.     

Using the EM algorithm for inference in a mixture of distributions with censored but partially identifiable data

This paper describes an implementation of the EM algorithm for the statistical analysis of a finite mixture of distributions arising when data are censored but partially identifiable. We consider a scheme of type I censoring where censoring times are random. The estimation of standard errors proposed by Meng and Rubin (1991. Using EM to obtain asymptotic variance-covariance matrices: the SEM algorithm. J. Amer. Statist. Assoc. 86(416), 899-909) is also implemented in the context of the above mixture. A Bayesian method introduced in Contreras-Cristán et al. (2003. Statistical inference for mixtures of distributions for censored data with partial identification. Commun. in Statist. Theory Methods 32(4), 749-774) for the case of a constant censoring value is extended to the case of random censoring times. Comparisons with different methods are carried out both with simulated data and with the observations on failure times for communication transmitter-receivers of Mendenhall and Hader (1958. Estimation of parameters of mixed exponentially distributed failure time distributions from censored life test data. Biometrika 45, 504-520).     

Implementing wavelet transform/mixture of experts network for analysis of electrocardiogram beats

Abstract: Mixture of experts (ME) is a modular neural network architecture for supervised learning. This paper illustrates the use of the ME network structure to guide model selection for classification of electrocardiogram (ECG) beats. The expectation maximization algorithm is used for training the ME so that the learning process is decoupled in a manner that fits well with the modular structure. The ECG signals were decomposed into time–frequency representations using discrete wavelet transforms and statistical features were calculated to depict their distribution. The ME network structure was implemented for ECG beats classification using the statistical features as inputs. To improve classification accuracy, the outputs of expert networks were combined by a gating network simultaneously trained in order to stochastically select the expert that is performing the best at solving the problem. Five types of ECG beats (normal beat, congestive heart failure beat, ventricular tachyarrhythmia beat, atrial fibrillation beat, partial epilepsy beat) obtained from the Physiobank database were classified with an accuracy of 96.89% by the ME network structure. The ME network structure achieved accuracy rates which were higher than those of the stand-alone neural network models.     

EEG signal classification using wavelet feature extraction and a mixture of expert model

Mixture of experts (ME) is modular neural network architecture for supervised learning. A double-loop Expectation-Maximization (EM) algorithm has been introduced to the ME network structure for detection of epileptic seizure. The detection of epileptiform discharges in the EEG is an important component in the diagnosis of epilepsy. EEG signals were decomposed into the frequency sub-bands using discrete wavelet transform (DWT). Then these sub-band frequencies were used as an input to a ME network with two discrete outputs: normal and epileptic. In order to improve accuracy, the outputs of expert networks were combined according to a set of local weights called the “gating function”. The invariant transformations of the ME probability density functions include the permutations of the expert labels and the translations of the parameters in the gating functions. The performance of the proposed model was evaluated in terms of classification accuracies and the results confirmed that the proposed ME network structure has some potential in detecting epileptic seizures. The ME network structure achieved accuracy rates which were higher than that of the stand-alone neural network model.     

空间约束层次加权Gamma混合模型的SAR图像分割

目的 合成孔径雷达（SAR）图像中像素强度统计分布呈现出复杂的特性,而传统混合模型难以建模非对称、重尾或多峰等特性的分布。为了准确建模SAR图像统计分布并得到高精度分割结果,本文提出一种利用空间约束层次加权Gamma混合模型（HWGaMM）的SAR图像分割算法。方法 采用Gamma分布的加权和定义混合组份;考虑到同质区域内像素强度的差异性和异质区域间像素强度的相似性,采用混合组份加权和定义HWGaMM结构。采用马尔可夫随机场（MRF）建模像素空间位置关系,利用中心像素及其邻域像素的后验概率定义混合权重以将像素邻域关系引入HWGaMM,构建空间约束HWGaMM,以降低SAR图像内固有斑点噪声的影响。提出算法结合M-H（Metropolis-Hastings）和期望最大化算法（EM）求解模型参数,以实现快速SAR图像分割。该求解方法避免了M-H算法效率低的缺陷,同时克服了EM算法难以求解Gamma分布中形状参数的问题。结果 采用3种传统混合模型分割算法作为对比算法进行分割实验。拟合直方图结果表明本文算法具有准确建模复杂统计分布的能力。在分割精度上,本文算法比基于高斯混合模型（GMM）、Gamma分布和Gamma混合模型（GaMM）分割算法分别提高33%,29%和9%。在分割时间上,本文算法虽然比GMM算法多64 s,但与基于Gamma分布和GaMM算法相比较分别快600 s和420 s。因此,本文算法比传统M-H算法的分割效率有很大的提高。结论 提出一种空间约束HWGaMM的SAR图像分割算法,实验结果表明提出的HWGaMM算法具有准确建模复杂统计分布的能力,且具有较高的精度和效率。     

特征点LBP信息在表情识别中的应用

提出一种基于特征点LBP信息的表情识别方法。在分析了表情识别中的LBP特征之后,选择含有丰富表情信息的上半脸眼部周围和下半脸嘴部周围的特征点,计算每个特征点邻域的LBP信息作为表情特征进行表情识别。实验表明,基于特征点LBP信息的方法不需要对人脸进行预配准,较传统的LBP特征更有利于表情识别的实现。     

Estimation for the number of components in a mixture model using stepwise split-and-merge EM algorithm

The main difficulty with EM algorithm for mixture model concerns the number of components, say g. This is the question of model selection, and the EM algorithm itself could not estimate g. On the contrary, the algorithm requires g to be specified before the remaining parameters can be estimated. To solve this problem, a new algorithm, which is called stepwise split-and-merge EM (SSMEM) algorithm, is proposed. The SSMEM algorithm alternately splits and merges components, estimating g and other parameters of components simultaneously. Also, two novel criteria are introduced to efficiently select the components for split or merge. Experimental results on simulated and real data demonstrate the effectivity of the proposed algorithm.     

Modeling,Reachability and Controllability of Bounded Petri Nets Based on Semi‐Tensor Product of Matrices

In this paper, we present a comprehensive modeling technique for bounded Petri net systems (BPNSs) in the framework of the semi‐tensor product (STP) of matrices. The two dynamic properties of BPNSs, namely, reachability and controllability, are investigated systematacially. First, the dynamics of a bounded Petri net system (BPNS), by resorting to the STP of matrices, are expressed in the form of a discrete‐time bilinear equation, which is called the marking evolution equation (MEE) of BPNSs. Second, controllability and transition‐marking adjacency matrix (TMAM) of BPNSs are defined, respectively. Further, several necessary and sufficient conditions for reachability and controllability of BPNSs are given in terms of the MEE and TMAM. Third, an efficient algorithm to verify reachability property of BPNSs, in this paper, is provided, as well as its computational complexity. Finally, an example is presented to illustrate the theoretical results in this paper. The main contribution of this paper is the presentation of a precise mathematical model for BPNSs. The main advantage of the proposed approach is that not only it can be applied to verify whether or not any given marking is reachable from the other in state space, but also it is very convenient to find all firing sequences between any two reachable markings.     

Dempster-Shafer证据融合金字塔韦伯局部特征的表情识别

目的 针对韦伯局部特征(WLD)在局部细节表征上的局限性,提出一种基于金字塔韦伯局部特征(PWLD)的人脸表情识别方法。方法 首先对人脸图像进行预处理,分割出表情显著区域;然后根据显著区域大小进行分层并加以分块,对每一层的不同块提取PWLD特征,将测试集与训练集的PWLD直方图的卡方距离归一化作为独立证据来构造基本概率分配(BPA);最后利用D-S(Dempster-Shafer)证据理论联合规则得到融合后的BPA,进而得出识别结果。结果 在JAFFE库与Cohn-Kanade库上进行交叉验证,平均识别率分别为95.81%和97.47%。结论 PWLD特征在局部细节表征上比WLD特征更具有优势。与目前典型的方法进行对比实验,验证该方法的有效性和容错性,同时证明该方法在噪声和光照条件下具有一定的鲁棒性。     

基于混合概率PCA模型高光谱图像本征维数确定

如何有效实现降维是现代成像光谱仪辨识地物类别的一个难点所在。该文在已知高光谱图像地物类别数的情况下，提出了一种采用混合最小描述长度(MMDL)模型选择准则确定高光谱图像本征维数的方法。该方法在期望最大化算法框架下同时实现混合PPCA降维和聚类，并根据MMDL准则确定数据降维维数，可以得到数据在概率意义下的精确的降维表征。仿真数据和真实数据进行的比较实验表明，该方法能精确地选择数据的本征维数。     

A new practical robust control of cable‐driven manipulators using time‐delay estimation

In this paper, a new practical robust control scheme is proposed and investigated for the cable‐driven manipulators under lumped uncertainties. There are three parts in the proposed method, ie, a time‐delay estimation (TDE) part, a modified super‐twisting algorithm (STA) part, and a fractional‐order nonsingular terminal sliding mode (FONTSM) error dynamics part. The TDE uses intentionally time‐delayed system signals to estimate the lumped dynamics of the system and ensures an attractive model‐free control structure. The STA is applied to guarantee high performance and chattering suppression simultaneously in the reaching phase. The FONTSM error dynamics is utilized to obtain fast convergence and strong robustness in the sliding mode phase. Thanks to the above three parts, the proposed control scheme is model free and can ensure high control performance under lumped uncertainties. The stability considering the FONTSM error dynamics and modified STA scheme is analyzed. Comparative simulation and experiments were conducted to demonstrate the effectiveness and superiorities of the newly proposed control scheme. Corresponding experimental results show that our newly proposed control scheme can provide more than 20% improvement of the steady control accuracy under three different reference trajectories.     

Set stabilization of boolean networks via sampled‐data control

The paper investigates the sampled‐data state feedback control (SDSFC) for set stabilization of Boolean control networks (BCNs). Set stabilization means that a system converges to a subset of the state space under certain controllers. Assume that the given subset is , where and sampling period is τ. We consider two conditions q ≤ τ;q > τ and for any given subset , calculate the corresponding largest control invariant subset (LCIS). Moreover, a design procedure to calculate all possible SDSFCs for set stabilization of BCNs is obtained. Ultimately, we provide an example to demonstrate the efficiency of the results.     

Fast solution of volume–surface integral equation for scattering from composite conducting‐dielectric targets using multilevel fast dipole method

This article presents a fast solution to the volume–surface integral equation for electromagnetic scattering from three‐dimensional (3D) targets comprising both conductors and dielectric materials by using the multilevel fast dipole method (MLFDM). This scheme is based on the concept of equivalent dipole‐moment method (EDM) that views the Rao–Wilton–Glisson and the Schaubert–Wilton–Glisson basis functions as dipole models with equivalent dipole moments. In the MLFDM, a simple Taylor's series expansion of the terms R^α (α = 1, ?1, ?2, ?3) and R? R? in the formulation of the EDM transforms the interaction between two equivalent dipoles into an aggregation–translation–disaggregation form naturally. Furthermore, benefiting from the multilevel grouping scheme, the matrix‐vector product can be accelerated and the memory cost is reduced remarkably. Simulation results are presented to demonstrate the efficiency and accuracy of this method. © 2012 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2012.     

Automatically solving two‐variable linear algebraic word problems using text mining

The teaching and learning of algebraic word problems is a basic component of elementary education. Recently, to facilitate its learning, a few approaches for automatically solving algebraic and arithmetic word problems have been proposed. These systems generally use either natural language processing (NLP) or a combination of NLP and machine learning. However, they have low accuracy due to their large feature sets, extracted using limited preprocessing techniques. In this research work, we propose a template‐based approach that was developed by following a two‐step process. In the first step, we predict an equation template from a training dataset using NLP and a classification mechanism. The next step is to instantiate the predicted template with nouns and numbers through reasoning. To validate the proposed methodology, a prototype system was implemented. We then compared the proposed system with the existing systems using their respective datasets and the proposed dataset. The experimental results show improvement in accuracy, with an average precision of 80.6% and average recall of 83.5%.     

Super‐Twisting Observer‐Based Integral Sliding Mode Control for Tracking the Rapid Acceleration of a Piston in a Hybrid Electro‐Hydraulic and Pneumatic System

A new hybrid electro‐hydraulic and pneumatic actuator system and its dynamic model for high‐performance control are presented. This work focuses on tracking control of rapidly changing acceleration that is an advanced area with various practical applications in industries. The impact motion control of the actuator is one of challenging task due to the system instability during the transition state. Since composite disturbances derived from the inaccurate and unmodeled dynamics considerably reduce the control performance. A novel structure of variable integral sliding mode controls integrated with a sliding mode disturbance observer is proposed based on the super‐twisting algorithm. With the control strategy, not only does the controller overcome the extreme sensitivity of the system during rapid movements, but it also eliminates the internal parameter uncertainties and external load disturbance while tracking rapid gain‐scheduled acceleration. The results of the numerical simulation and field experiment are presented to assess the effectiveness of the proposed control scheme.     

Analysis and improvement of PSRR in an envelope‐tracking power amplifier using a common‐drain amplifier topology

A common‐drain power amplifier (PA) for envelope tracking systems is presented. In envelope tracking, the main PA operates mostly in compression and the power supply rejection ratio (PSRR) is not high. Furthermore, the output noise of the supply modulator can be mixed with the RF signal and generates out‐of‐band emissions. In this article, instead of using a common‐source topology, the PSRR of the envelope tracking PA is inherently improved by utilizing a common‐drain topology. A comprehensive analysis shows that the common‐drain topology is less sensitive to the supply noise, as compared to the conventional common‐source topology. The proposed common‐drain PA is implemented using a discrete LDMOS PD20010‐E RF transistor. Measurement results show that the PSRR of the proposed common‐drain PA is improved by up to 7 dB as compared to that of the common‐source PA. For a two‐tone input with 10 MHz bandwidth at the center frequency of 700 MHz, the power added efficiency (PAE) and IM3 of the envelope tracking common‐drain PA are 20% and ? 28 dBc, respectively, at an average output power of 33.4 dBm. The amplifier also shows a 12.4 dB power gain. Moreover, by utilizing the envelope tracking, the PAE is improved by more than 5%.     

L2 stability,H∞ control of switched homogeneous nonlinear systems and their semi‐tensor product of matrices representation

This paper investigates the problems of L₂ stability and H_∞ control of switched homogeneous nonlinear systems. This paper gives that if a switched homogeneous nonlinear system with disturbance is internally homogeneously asymptotically stable, then it has a finite L₂ gain, and if a switched homogeneous nonlinear system with disturbance and controls is homogeneously stabilizable under the zero disturbance condition, its H_∞ control problem is solvable under some mild conditions, and a homogeneous solution is given. Then, via the semi‐tensor product of matrices method, the aforementioned obtained results are transformed to linear‐like forms, and the aforementioned obtained Hamilton–Jacobi–Isaacs inequality is transformed to a linear‐like matrix inequality, which makes it feasible to compute the solutions by computer. Copyright © 2012 John Wiley & Sons, Ltd.     

Performance analysis of 2‐DOF tracking control for a class of nonlinear uncertain systems with discontinuous disturbances

In this paper, the tracking control problem is considered for a class of uncertain nonlinear systems with infinite discontinuous points in the external disturbance. The extended state observer–based 2‐degree‐of‐freedom control is used with one degree to estimate and cancel the “total disturbance” and the other to force the closed‐loop system to have desired characteristics. The tracking error between the state vector and its ideal trajectory in the entire transient process is adequately discussed to illuminate the performance of resulting control systems. The quantitative analysis shows that the tracking error can be small enough by tuning the bandwidth of the extended state observer. Moreover, the necessary and sufficient condition for the tracking error and the estimation error of the “total disturbance” to converge to zero is presented. The simulation results of a motion test demonstrate that the desired performance of the control system can be achieved despite discontinuous disturbance and nonlinear uncertainties.     

Performance of a circular crossed‐dipole array for SDMA configuration adopting directivity and polarization control using particle swarm optimization algorithm

In this article, the performance of a circular crossed‐dipole array (CCDA) for space division multiple access (SDMA) configurations adopting directivity and polarization control is presented. The array consists of 12 dual‐polarized elements uniformly distributed in a circular configuration; each dual‐polarized element (crossed‐dipole) consists of two half‐wave dipoles in a ±45° slant configuration. The modified particle swarm optimization and moment of method (MPSO‐MOM) algorithm is used to calculate the complex weightings of the array elements in a mutual coupling environment for beamforming synthesis. In addition, the performance of the adaptive array using discrete feedings (1‐bit amplitude and 4‐bit phase shifters or only 4‐bit phase shifters) is studied. © 2008 Wiley Periodicals, Inc. Int J RF and Microwave CAE, 2009.     

Design and prototype manufacturing of a feed system for Ku‐band satellite communication by using 3D FDM/PLA printing and conductive paint technology

In this study, the realization of a Ku‐band feed system for reflector antenna in satellite communication systems is presented using 3D printing and conductive paint methods. The system includes a corrugated conical horn antenna designed to operate at 10.5 to 18.5 GHz and an H‐plane waveguide diplexer to operate at 10.7 to 12.75 GHz and 17.3 to 18.4 GHz in receive (RX) and transmit (TX) bands, respectively. In the manufacturing of the structures, fused deposition modeling (FDM) technology and polylactic acid material are processed for 3D printing, where nickel and silver conductive‐based paints are used for coating purpose. The measurement results of the feed system are found to be in good agreement with simulations that the combined (nickel‐coated antenna and silver‐coated diplexer) structure has return loss of more than 10 dB and high gain performance of 12 to 17 dBi within the RX and TX bands of 10.7 to 12.75 GHz and 17.4 to 18.8 GHz, respectively; while rejection (isolation) level between TX and RX ports is higher than 60 dB. The complex structure containing several detailed shapes inside shows that this low‐cost production technique as compared to high‐cost CNC‐based metallic production technology can be used for the prototype structures or proof‐of‐concept type studies of Ku‐band systems.