A divide-and-conquer-method for the synthesis of liveness enforcing supervisors for flexible manufacturing systems

In this paper a divide-and-conquer-method for the synthesis of liveness enforcing supervisors (LES) for flexible manufacturing systems (FMS) is proposed. Given the Petri net model (PNM) of an FMS prone to deadlocks, it aims to synthesize a live controlled Petri net system. For complex systems, the use of reachability graph (RG) based deadlock prevention methods is a challenging problem, as the RG of a PNM easily becomes unmanageable. To obtain the LESs from a large PNM is usually intractable. In this paper, to ease this problem the PNM of a system is divided into small connected subnets. Each connected subnet prone to deadlocks is then used to compute the LES for the original PNM. Starting from the simplest subnet prone to deadlocks to make the subnet live, monitors (control places) are computed. The RG of each subnet is considered and split into a dead-zone (DZ) and a live-zone. All states in the DZ are prevented from being reached by means of a well-established invariant-based control method. Next, the computation of monitors is followed for bigger subnets. Previously computed monitors are included within the bigger subnets based on a criterion. This process keeps the DZ of the bigger subnets smaller compared with the original uncontrolled subnets. When all subnets are live we obtain a set of monitors that are included within the PNM to obtain a partially controlled PNM (pCPNM). A new set of monitors is also computed for the pCPNM. Finally, a live controlled Petri net system is obtained. The proposed method is generally applicable, easy to use, effective and straightforward although its off-line computation is of exponential complexity in theory. Its use for FMS control guarantees deadlock-free operation and high performance in terms of resource utilization and system throughput. Two FMS deadlock problems from the literature are used to illustrate the applicability and the effectiveness of the proposed method.     

Gradually tapered light pipes for illumination of LED projectors

Abstract— LEDs are totally different from classical light sources: they have a different shape, radiation pattern, driving requirements, etc. Therefore, the illumination engine, which determines the brightness and uniformity of the system, has to be redesigned for LED‐based projectors. A compact illumination system based on gradually tapered light pipes (GTLP) will be presented. The GTLP collects, reshapes, and uniformizes the light flux from the LED to illuminate the light‐valve uniformly. The design and the simulations have been completed. The result is a uniformly illuminated rectangular beam at the end of the pipe with an efficiency of 81.1%. Afterwards, the light pipe was fabricated, and the experimentally measured efficiency is 76.9%, which demonstrates a successful manufacturing process. Finally, two recycling techniques to enhance the brightness have been applied and these enhancements were experimentally observed.     

Acoustic modes of self-excited oscillations in struts

The acoustic modes of self-excited oscillations are investigated numerically for two-dimensional generic struts. In profiles with short chord lengths, hydrodynamic instability is responsible for wake bifurcation and for generation of narrow band tones. However, it is found, another mechanism is the major factor for periodic oscillations in long profiles. Two points of wall contour develops an acoustic feedback between each other. These points are the inflection point and the trailing edge point. Consequently, limits of two different modes of periodic oscillations are outlined.Moreover, it is determined that acoustic radiation has dipole character which agrees well with the theory of sound radiation for acoustically compact surfaces.     

Equilibrium molecular dynamics studies on nanoscale-confined fluids

Fluid behavior within nanoscale confinements is studied for argon in dilute gas, dense gas, and liquid states. Molecular dynamics simulations are used to resolve the density and stress variations within the static fluid. Normal stress calculations are based on the Irving–Kirkwood method, which divides the stress tensor into its kinetic and virial parts. The kinetic component recovers pressure based on the ideal-gas law. The particle–particle virial increases with increased density, whereas the surface–particle virial develops because of the surface-force field effects. Normal stresses within nanoscale confinements show anisotropy primarily induced by the surface-force field and local variations in the fluid density near the surfaces. For dilute and dense gas cases, surface-force field that extends typically 1 nm from each wall induces anisotropic normal stress. For liquid case, this effect is further amplified by the density fluctuations that extend beyond the force field penetration region. Outside the wall-force field penetration and density fluctuation regions, the normal stress becomes isotropic and recovers the thermodynamic pressure, provided that sufficiently large force cut-off distances are used in the computations.     

Condition monitoring of speed controlled induction motors using wavelet packets and discriminant analysis

This work presents an intelligent method for the condition monitoring of induction motors supplied with adjustable speed drives (ASD). Most of the previous work in this area concentrated on the fault detection and classification of induction motors supplied directly from an a.c. line. However, ASD driven induction motors are widely used in industrial processes and therefore obtaining an intelligent tool for the condition monitoring of these motors is necessary in terms of preventive maintenance and reducing down time due to motor faults. Here 3-phase supply side current of the ASD driving an induction motor is used to extract statistical features of wavelet packet decomposition coefficients within a frequency range of interest. This way, the information regarding the output frequency of the ASD and hence the motor speed is not required. Six identical three-phase induction motors were used for the experimental verification of the proposed method. One healthy machine was used as a reference, while other five with various synthetic faults were used for condition detection and classification. Extracted features obtained from decomposition coefficients of different wavelet filter types for all motors were employed in three different and popular classifiers. The proposed method and the performance of the features used for fault detection and classification are examined at various motor loads and speed levels, and it is shown that a successful condition monitoring system for induction motors supplied with ASDs is developed. The effect of selected filter type in wavelet decomposition to the condition monitoring process is analyzed and presented.     

Higher-Order Smoothing： A Novel Semantic Smoothing Method for Text Classification

It is known that latent semantic indexing （LSI） takes advantage of implicit higher-order （or latent） structure in the association of terms and documents. Higher-order relations in LSI capture ＂latent semantics＂. These findings have inspired a novel Bayesian framework for classification named Higher-Order Naive Bayes （HONB）, which was introduced previously, that can explicitly make use of these higher-order relations. In this paper, we present a novel semantic smoothing method named Higher-Order Smoothing （HOS） for the Naive Bayes algorithm. HOS is built on a similar graph based data representation of the HONB which allows semantics in higher-order paths to be exploited. We take the concept one step further in HOS and exploit the relationships between instances of different classes. As a result, we move beyond not only instance boundaries, but also class boundaries to exploit the latent information in higher-order paths. This approach improves the parameter estimation when dealing with insufficient labeled data. Results of our extensive experiments demonstrate the value of HOS oi1 several benchmark datasets.     

Surface charge-dependent transport of water in graphene nano-channels

Deionized water flow through positively charged graphene nano-channels is investigated using molecular dynamics simulations as a function of the surface charge density. Due to the net electric charge, Ewald summation algorithm cannot be used for modeling long-range Coulomb interactions. Instead, the cutoff distance used for Coulomb forces is systematically increased until the density distribution and orientation of water atoms converged to a unified profile. Liquid density near the walls increases with increased surface charge density, and the water molecules reorient their dipoles with oxygen atoms facing the positively charged surfaces. This effect weakens away from the charged surfaces. Force-driven water flows in graphene nano-channels exhibit slip lengths over 60 nm, which result in plug-like velocity profiles in sufficiently small nano-channels. With increased surface charge density, the slip length decreases and the apparent viscosity of water increases, leading to parabolic velocity profiles and decreased flow rates. Results of this study are relevant for water desalination applications, where optimization of the surface charge for ion removal with maximum flow rate is desired.     

An expert system based on S-transform and neural network for automatic classification of power quality disturbances

In this paper, an S-transform-based neural network structure is presented for automatic classification of power quality disturbances. The S-transform (ST) technique is integrated with neural network (NN) model with multi-layer perceptron to construct the classifier. Firstly, the performance of ST is shown for detecting and localizing the disturbances by visual inspection. Then, ST technique is used to extract the significant features of distorted signal. In addition, an optimum combination of the most useful features is identified for increasing the accuracy of classification. Features extracted by using the S-transform are applied as input to NN for automatic classification of the power quality (PQ) disturbances that solves a relatively complex problem. Six single disturbances and two complex disturbances as well pure sine (normal) selected as reference are considered for the classification. Sensitivity of proposed expert system under different noise conditions is investigated. The analysis and results show that the classifier can effectively classify different PQ disturbances.     

基于多策略的维吾尔文网页识别方法

经过对大量维吾尔文网站的调查与分析,该文从多语种混合网页中针对维吾尔文网页识别进行了研究,这对维吾尔语信息处理工作起着关键作用。首先该文探讨了维吾尔文不规范网页的字符编码转换规则及原理,以此对不规范维吾尔文字符进行了相应的处理,之后介绍了基于修改的N-Gram方法和基于维吾尔语常用词特征向量的两种方法,其中后者融合了维吾尔文常用候选词语料库及向量空间模型(Vector Space Model)。使用三种不同类型的维吾尔文网页文本作为本研究的数据集,在此基础上验证了该文提出的网页识别方法,以及采用不同的方法进行了网页识别的实验。实验结果表明,基于N-Gram的方法对正文较长的新闻或论坛网页的识别性能最佳,反而基于常用词特征向量的方法对短文本的网页识别性能优越N-Gram。所提方法对维吾尔文网页识别的整体性能达到90%以上,并验证了这两种方法的有效性。