A maximum a posteriori super resolution algorithm based on multidimensional Lorentzian distribution

This paper presents a threshold-free maximum a posteriori （MAP） super resolution （SR） algorithm to reconstruct high resolution （HR） images with sharp edges. The joint distribution of directional edge images is modeled as a multidimensional Lorentzian （MDL） function and regarded as a new image prior. This model makes full use of gradient information to restrict the solution space and yields an edge-preserving SR algorithm. The Lorentzian parameters in the cost function are replaced with a tunable variable, and graduated nonconvexity （GNC） optimization is used to guarantee that the proposed multidimensional Lor- entzian SR （MDLSR） algorithm converges to the global minimum. Simulation results show the effectiveness of the MDLSR algorithm as well as its superiority over conventional SR methods.     

Rate-equation-based VCSEL thermal model and simulation

INTRODUCTION Vertical-Cavity Surface-Emitting Lasers (VCSELs) are a new generation of semiconductor lasers that differ considerably from the conventional edge-emitting lasers. In recent years, the characteris- tics of VCSELs have improved enormously (Badilita et al., 2004). Today’s VCSELs have low-threshold current densities and high output power. Moreover, their circular output beam profiles and the suitability to be integrated into 2D arrays make them most promising candidates …     

对话、念读两种说话状态下声纹特征变化的实验研究

在对话、念读两种说话状态下,对20位青年男性普通话语流中240个音段所包含的单元音[a]、[i]、[u]的共鸣方式、平均音高、音强等听觉特征,共振峰走向、强度、频率等频谱特征进行定性和定量分析。发现说话状态改变后个人的声纹特征有着不同程度的变化:听觉特征、共振峰强度特征的变化较大,共振峰走向和频率特征相对稳定。在被检测的音段中,40.0%的音段听感音高和音强发生了变化,45.3%的音段共振峰强度发生了变化;尽管有13.3%的音段共振峰走向存在差异,但共振峰频率变异系数的变化范围均在2.7%以内。由此可见,在对不同说话状态下的语音进行声纹鉴定时,应重点关注共振峰走向和频率,以判断涉案语音和嫌疑人样本语音是否同一。     

模糊神经网络变结构算法优化的研究

随着生产实际情况的不断变化,以及模糊神经网络不断的改进和发展,提出一种改进的构造神经网络的方法,并且提出混合学习算法,结合共扼梯度下降法与递归最小二乘估计来分别辨识网络中的前、后件参数,并对非线性系统进行仿真实验,达到控制要求.     

Pithc-synchronous articulatory synthesis incorporated with the inverse solution of speech production

This paper presents a new proposal to synthesize natural sounds with less control parameters by combining the inverse speech production and pitch-synchronous articulatory synthesis. The pitch-synchronous excited Reflection-Type Line Analog (RTLA) model is employed as the synthesis filter. Multi-rate system sampling and dynamic scattering wave adjustment are used to handle the variable VT length and the acoustic continuity. The synthesizer is controlled by vocal-tract (VT) area functions. Given the targets of formant trajectories, the dynamic VT area function which is modeled by time variant VT length is derived using an inverse solution of speech production. A distinguishing feature of this method is that artificially specified formant trace can be precisely aimed in the synthetical sounds. Experimental results show that the formant target can be well matched by the synthetic sounds. Potential application to text-to-speech conversion of this method is discussed. Project supported by NSFC (69972046), and Zhenjiang Provincial Natural Science Foundation of China (698076).     

New hybrid model of proton exchange membrane fuel cell

Model and simulation are good tools for design optimization of fuel cell systems. This paper proposes a new hybrid model of proton exchange membrane fuel cell (PEMFC). The hybrid model includes physical component and black-box com-ponent. The physical component represents the well-known part of PEMFC, while artificial neural network (ANN) component estimates the poorly known part of PEMFC. The ANN model can compensate the performance of the physical model. This hybrid model is implemented on Matlab/Simulink software. The hybrid model shows better accuracy than that of the physical model and ANN model. Simulation results suggest that the hybrid model can be used as a suitable and accurate model for PEMFC.     

PITHC-SYNCHRONOUS ARTICULATORY SYNTHESIS INCORPORATED WITH THE INVERSE SOLUTION OF SPEECH PRODUCTION

ＩＮＴＲＯＤＵＣＴＩＯＮＳｙｎｔｈｅｓｉｓｏｆｎａｔｕｒａｌｓｐｅｅｃｈｓｏｕｎｄｓｗｉｔｈｌｅｓｓｃｏｎｔｒｏｌｐａｒａｍｅｔｅｒｓｈａｓｐｒａｃｔｉｃａｌｓｉｇｎｉｆｉｃａｎｃｅ ,ｅｓ ｐｅｃｉａｌｌｙｉｎｔｅｘｔ ｔｏ ｓｐｅｅｃｈ (ＴＴＳ)ｆｏｒｐｒｏｄｕｃｉｎｇｓｏｕｎｄｓｗｉｔｈａｒｔｉｆｉｃｉａｌｌｙｓｐｅｃｉｆｉｅｄｆｏｒｍａｎｔｔａｒｇｅｔｓａｎｄｓｃａｌｅｄｐｉｔｃｈｐａｒａｍｅｔｅｒｓｏｔｈａｔｔｈｅｔｉｍｂｒｅｏｆｓｐｅｅｃｈｏｕｔｐｕｔｃａｎｂｅａｒｂｉｔｒａｒｉｌｙｔｏｎ…     

Extracting invariable fault features of rotating machines with multi-ICA networks

This paper proposes novel multi-layer neural networks based on Independent Component Analysis for feature extraction of fault modes. By the use of ICA, invariable features embedded in multi-channel vibration measurements under different operating conditions (rotating speed and/or load) can be captured together. Thus, stable MLP classifiers insensitive to the variation of operation conditions are constructed. The successful results achieved by selected experiments indicate great potential of ICA in health condition monitoring of rotating machines. Project supported by the National Natural Science Foundation (No. 50205025) and the Natural Science Foundation of Zhejiang Province (No. 5001004).     

Extracting invariable fault features of rotating machines with multi-ICA networks

INTRODUCTIONRotatingmachinessuchaselectromotor,dy namotor,turbocompressor,etc .areimportantequipmentsinmanyindustryfields.Peoplehadbeenpayingconsiderableattentiontotheircon ditionmonitoringandfaultdiagnosis (Xu ,19 98) .Theoretically ,anyfluctuationofforc…     

Nonlinear modelling of a SOFC stack by improved neural networks identification

The solid oxide fuel cell (SOFC) is a nonlinear system that is hard to model by conventional methods. So far,most existing models are based on conversion laws,which are too complicated to be applied to design a control system. To facilitate a valid control strategy design,this paper tries to avoid the internal complexities and presents a modelling study of SOFC per-formance by using a radial basis function (RBF) neural network based on a genetic algorithm (GA). During the process of mod-elling,the GA aims to optimize the parameters of RBF neural networks and the optimum values are regarded as the initial values of the RBF neural network parameters. The validity and accuracy of modelling are tested by simulations,whose results reveal that it is feasible to establish the model of SOFC stack by using RBF neural networks identification based on the GA. Furthermore,it is possible to design an online controller of a SOFC stack based on this GA-RBF neural network identification model.     

面向侧面的软件逆向工程

1IntroductionSoftware reverse engineering is a process of analyz-ing a software system to identify its components andtheir interrelationships and create representations ofthe system in another form,or at a higher level ofabstraction[1].Xi Dian reverse engineering tool(XDRE)[2]previously developed in the author’s labcan partially recover the UML diagrams from C source codes[3].However,the fact that crosscuttingconcerns cannot be extracted by most software reverseengineeringtools is ani mp…     

Parameters optimization and nonlinearity analysis of grating eddy current displacement sensor using neural network and genetic algorithm

A grating eddy current displacement sensor （GECDS） can be used in a watertight electronic transducer to realize long range displacement or position measurement with high accuracy in difficult industry conditions. The parameters optimization of the sensor is essential for economic and efficient production. This paper proposes a method to combine an artificial neural network （ANN） and a genetic algorithm （GA） for the sensor parameters optimization. A neural network model is developed to map the complex relationship between design parameters and the nonlinearity error of the GECDS, and then a GA is used in the optimization process to determine the design parameter values, resulting in a desired minimal nonlinearity error of about 0.11%. The calculated nonlinearity error is 0.25%. These results show that the proposed method performs well for the parameters optimization of the GECDS.     

Optimal choice of parameters for particle swarm optimization

INTRODUCTION Particle swarm optimization (PSO) is one of the evolutionary computational techniques. Since its introduction (Kennedy and Eberhart, 1995; Eberhart and Kennedy, 1995; Eberhart et al., 1996), PSO has attracted much attention from researchers around the world. It is a population-based search algorithm and is initialized with a population of random solutions, called particles. Each particle in PSO moves over the search space at velocity dynamically adjusted according to th…     

Shape optimization of plate with static and dynamic constraints via virtual laminated element

The virtual laminated element method (VLEM) can resolve structural shape optimization problems with a new method. According to the characteristics of VLEM, only some characterized layer thickness values need be defined as design variables instead of boundary node coordinates or some other parameters determining the system boundary. One of the important features of this method is that it is not necessary to regenerate the FE (finite element) grid during the optimization process so as to avoid optimization failures resulting from some distortion grid elements. The thickness distribution in thin plate optimization problems in other studies before is of stepped shape. However, in this paper, a continuous thickness distribution can be obtained after optimization using VLEM, and is more reasonable. Furthermore, an approximate reanalysis method named “behavior model technique” can be used to reduce the amount of structural reanalysis. Some typical examples are offered to prove the effectiveness and practicality of the proposed method. Project (No. 50075083) supported by National Natural Science Foundation of China     

情感语音信息中共振峰参数的提取方法

共振峰参数是反映声道特性的一个重要参数，不同情感的发音使声道有不同的变化，所以不同情感发音的共振峰位置不同，首先采用线性预测法求出预测系数，然后用预测系数估计声道的功率谱，再用峰值检出法算出共振峰频率，运用此方法，可以判别出不同情感信息的共振峰频率，从而实现语音的情感识别。     

基于WEB的快速成型数据智能化处理技术的初步研究

Representations of contemporary rapid prototyping (RP) technologies may not be able to manage and share data processing information based on network. This paper presents a new approach, together with its technical implementation, to solve this problem, and discusses web-based study on RP data processing tecihniques. At present STL files produced by 3D modeling systems contain triangular facet representation of surfaces and have become standard data inputs of RP and manufacturing systems. The paper proposes algorithms for correcting common errors, and optimizes the algorithm of gaps or hole errors leading to open loops in the data structure based on analyses of the rules and errors in the STL file. In addition, effectiveness of the approach is demonstrated by a simple model view example of STL files based on Java 3D technology.     

Banach空间中的局部各向一致凸和局部弱紧一致凸(英文)

1977年,Mark.A.Simith在其博土论文中(见[6])引入Banach空间中的弱紧、一致凸(URWC)凸性新概念,并给出了一个Banach空间是URED但不是URWC的例子,以及一个Banach空间是URWC但不是WUR的例子。本文继续Mark.A的工作,引入了Banach空间中的局部各向一致凸(LUREO)和局部弱紧一致凸(LURWC)两个凸性新概念。作为本文的重要结果,本文给出了一个Banach空间是LURED但既不是URED也不是LURWC的例子;一个Banach空间是LURWC但不是LURWC的例子。特别,本文还证明了熟知的严格凸(SR),实际上闵是本文所新引入的一种局部一致凸性,即:局部各向一致凸。从而使严格凸在凸性分类中找到了一个恰当的位置。通过本文的工作,使Banach空间中的凸性类型得到了完善。     

Extraction of novel features for emotion recognition

Hilbert-Huang transform method has been widely utilized from its inception because of the superiority in varieties of areas. The Hilbert spectrum thus obtained is able to reflect the distribution of the signal energy in a number of scales accurately. In this paper, a novel feature called ECC is proposed via feature extraction of the Hilbert energy spectrum which describes the distribution of the instantaneous energy. The experimental results conspicuously demonstrate that ECC outperforms the traditional short-term average energy. Combination of the ECC with mel frequency cepstral coefficients (MFCC) delineates the distribution of energy in the time domain and frequency domain, and the features of this group achieve a better recognition effect compared with the feature combination of the short-term average energy, pitch and MFCC. Afterwards, further improvements of ECC are developed. TECC is gained by combining ECC with the teager energy operator, and EFCC is obtained by introducing the instantaneous frequency to the energy. In the experiments, seven status of emotion are selected to be recognized and the highest recognition rate 83.57% is achieved within the classification accuracy of boredom reaching 100%. The numerical results indicate that the proposed features ECC, TECC and EFCC can improve the performance of speech emotion recognition substantially.     

Support Vector Machine for mechanical faults classification

INTRODUCTION Support Vector Machine (SVM) is a relativelynew soft computing method based on statisticallearning theory presented by Vapnik (1995). In SVM,original input space is mapped into a high dimen-sional dot product space called feature space in whichthe optimal hyperplane is determined to maximize thegeneralization ability of the classifier. The optimalhyperplane is found by exploiting a branch ofmathematics, called optimization theory, and re-specting the insights provided by …     

一种深层过程神经网络及其在信号分类中的应用

针对时变信号小样本集建模分类问题,提出一种深层多尺度径向基过程神经网络（DLMS-RBFPNN）。该模型由时变信号输入层、多尺度径向基核变换层、全连接层和感知机分类器构成。兼顾时变信号的频谱特征和分布形态的多样性,基于径向基过程神经网络,通过将不同宽度参数的Gauss核函数进行线性叠加,构成多尺度核,完成不同尺度上对过程信号形态特征的提取、辨识和相似性度量。通过在径向基核函数层之上叠加全连接层和分类器,实现时变信号不同尺度特征的融合和分类。DLMS-RBFPNN具有较少的模型参数,适用于小样本集建模,在机制上可提高对时变信号过程细节特征和趋势特征的辨识及记忆能力。在分析DLMS-RBFPNN性质的基础上,建立一种基于动态聚类算法的核中心函数确定方法以及基于PSO的模型参数优化求解算法。以旋转机械基于示功图信号的故障诊断为例进行实验,结果验证了模型和算法的有效性。