Waveband-switchable SOA ring laser constructed with a phase modulator loop mirror filter

An electrically tunable multiwavelength source has been developed using a birefringence-based optical comb filter in a semiconductor optical amplifier ring laser. The filter is constructed with an electrooptic phase modulator placed inside a fiber loop mirror. By controlling the birefringence of the modulator through the applied bias, the radio-frequency power, or the modulation frequency, we achieve a continuous shift of the spectral comb to access different interleaved wavebands. Electrical waveband switching has been successfully demonstrated for 21 wavelengths at 100-GHz grid spacing with an optical signal-to-noise ratio over 40 dB.     

Double cracking in brittle cylindrical specimens subjected to quenching

A series of quench tests on hollow cylindrical specimens made of the machinable ceramic, Pyrophyllite, has been performed. The specimens consistently fractured with two diametrically opposite longitudinal cracks. Finite element analysis indicates that the primary crack occurred at the stage when the strain energy of the cylinder reached its maximum during the thermal transient process, while the secondary crack was caused by stress waves generated following the primary crack. This is supported by experimental observation of the time of fracture and the crack morphology. The work reported here helps to shed light on the fracture criterion for brittle cylindrical components under transient thermal and dynamic loadings.     

Double blind, randomised, placebo controlled study of oral vancomycin in prevention of necrotising enterocolitis in preterm, very low birthweight infants

AIMS: To evaluate the effectiveness of oral vancomycin in the prophylaxis of necrotising enterocolitis in preterm, very low birthweight infants. METHODS: A prospective, double blind, randomised, placebo controlled study in a tertiary referral centre of a university teaching hospital was conducted on 140 very low birthweight infants consecutively admitted to the neonatal unit. The babies were randomly allocated to receive oral vancomycin (15 mg/kg every 8 hours for 7 days) or an equivalent volume of placebo solution. Prophylaxis was started 24 hours before the start of oral feeds. All suspected cases of necrotising enterocolitis were investigated with a full sepsis screen and serial abdominal radiographs. Necrotising enterocolitis was diagnosed and staged according to modified Bell's criteria. RESULTS: Nine of 71 infants receiving oral vancomycin and 19 of 69 infants receiving the placebo solution developed necrotising enterocolitis (p = 0.035). Infants with necrotising enterocolitis were associated with a significant increase in mortality (p = 0.026) and longer duration of hospital stay (p = 0.002). CONCLUSIONS: Prophylactic oral vancomycin conferred protection against necrotising enterocolitis in preterm, very low birthweight infants and was associated with a 50% reduction in the incidence. However, widespread implementation of this preventive measure is not recommended, as it would only be effective in necrotising enterocolitis caused by Gram positive organisms and could increase the danger of the emergence of vancomycin resistant or dependent organisms. Its use should be restricted to a high prevalence nursery for a short and well defined period in a selected group of high risk patients.     

Automated analysis of nerve-cell images using active contour models

The number of nerve fibers (axons) in a nerve, the axon size, and shape can all be important neuroanatomical features in understanding different aspects of nerves in the brain. However, the number of axons in a nerve is typically in the order of tens of thousands and a study of a particular aspect of the nerve often involves many nerves. Potentially meaningful studies are often prohibited by the huge number involved when manual measurements have to be employed. A method that automates the analysis of axons from electron-micrographic images is presented. It begins with a rough identification of all the axon centers by use of an elliptical Hough transform procedure. Boundaries of each axons are then extracted based on active contour model, or snakes, approach where physical properties of the axons and the given image data are used in an optimization scheme to guide the snakes to converge to axon boundaries for accurate sheath measurement. However, false axon detection is still common due to poor image quality and the presence of other irrelevant cell features, thus a conflict resolution scheme is developed to eliminate false axons to further improve the performance of detection. The developed method has been tested on a number of nerve images and its results are presented.     

Integrated knowledge-based Petri net intelligent flexible assembly planning

Automatic assembly planning is recognized as an important tool for reducing manufacturing costs in concurrent product and process development. A novel knowledge-based Petri net (KBPN) is defined, based on the incorporation of expert systems into the usual Petri nets, and used for a unified assembly knowledge representation scheme. A KBPN-approach integrated with a sequence generation algorithm is proposed for the modeling, planning, simulation, analysis and evaluation of the flexible assembly system (FAS). The developed KBPN-based assembly planning system (KAPS) can automatically adjust the deviations between the theoretical planning parameters and the process parameters of real assembly operations to guarantee the best strategies and plans (sequences) for flexible assembly. The research findings are exemplified with a simple assembly to show the effectiveness of the method.     

A rough set approach to design concept analysis in a design chain

The inherent dynamic relationships among design tasks performed concurrently at different organizations characterize the complexities of a design chain where designers with diverse expertise need to collaborate across organizational boundaries. To ensure timely completion of inter-related design tasks, metrics to facilitate the early evaluation of design concepts are crucial. The ability to evaluate and select suitable design concepts at an early stage will ensure better solutions and greater savings in time and effort further downstream. This paper proposes a new approach based on the rough set theory to design concept analysis. The approach aims at early detection of design inadequacy. A so-called information system is constructed using the information gleaned from design concepts and design capabilities, and analyzed using the rough set theory to derive a set of design rules for design concept analysis. The approach embodies a technique for handling attributes with unavailable information, which is a frequent occurrence in design. This paper presents details of the proposed approach, the novel technique, and a case study.     

Untersuchungen zur Spin-Trapping-Kinetik in Systemen von Radikalen unterschiedlicher Reaktivität. Die Photolyse von α-Phenylbenzoin in Gegenwart von Benzyliden-tert.-butylamin-N-oxid

Investigations on Spin-Trapping Kineties in Radical Systems with Different Reactivity. Photolyses of α-Phenylbenzoine in the Presence of Benzylideneter. -butylamine N-oxide As a model system to study the spin-trapping kinetics of systems forming two radicals of different reactivity α-phenylbenzoine (PhB) was chosen. The photolysis of PhB forms benzoyl (B·) and diphenylketylradikals (K·), respectively, which can be trapped by benzylidene-tert-butyl-amine N-oxide (PBN), as a spin trap. The rate constants of radical addition to PBN and their reactions with the spin adducts were determined by means of flash photolysis and kinetic calculations of the PBN concentration dependence. For B· and K· addtion rate constants to PBN of k₁ = 8,1 × 10⁵ M⁻¹s⁻¹, and k₂ = 5,8 · 10⁴ M⁻¹s⁻¹, respectively, were found. The rate constants for the reactions of B· and K· with the spin adducts are in the order of 10⁹…10¹⁰ M⁻¹s⁻¹.     

Conceptual Design Using Evolution Strategy

Conceptual design is one of the most important activities in new product development. Inappropriate design concepts could result in high redesign cost and delay in product realisation. The generation of conceptual designs is a time-consuming and mentally exhausting process. It involves combining design features to generate as many potential design concepts as possible. As not all features combinations are feasible, decision-makers have to narrow down the potential solutions and subsequently select appropriate concept(s) for further development. This work concerns the automation of the conceptual design process using an evolutionary strategy. The paper proposes a design attribute encapsulation method, which is applied to the chromosome representation of the design concept, to facilitate the search for a promising combination of features. The design of golf clubs for children is used to illustrate the application of the proposed approach. Mathematical analysis shows that with the proposed approach, the probability of generating a feasible concept during the mutation and crossover is higher with better use of the feasibility rules. ID="A1" Correspondence and offprint requests to: S. C. Fok, School of Mechanical and Production Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798. E-mail: mscfok@ntu.edu.sg     

Efficient training algorithms for a class of shunting inhibitory convolutional neural networks

This article presents some efficient training algorithms, based on first-order, second-order, and conjugate gradient optimization methods, for a class of convolutional neural networks (CoNNs), known as shunting inhibitory convolution neural networks. Furthermore, a new hybrid method is proposed, which is derived from the principles of Quickprop, Rprop, SuperSAB, and least squares (LS). Experimental results show that the new hybrid method can perform as well as the Levenberg-Marquardt (LM) algorithm, but at a much lower computational cost and less memory storage. For comparison sake, the visual pattern recognition task of face/nonface discrimination is chosen as a classification problem to evaluate the performance of the training algorithms. Sixteen training algorithms are implemented for the three different variants of the proposed CoNN architecture: binary-, Toeplitz- and fully connected architectures. All implemented algorithms can train the three network architectures successfully, but their convergence speed vary markedly. In particular, the combination of LS with the new hybrid method and LS with the LM method achieve the best convergence rates in terms of number of training epochs. In addition, the classification accuracies of all three architectures are assessed using ten-fold cross validation. The results show that the binary- and Toeplitz-connected architectures outperform slightly the fully connected architecture: the lowest error rates across all training algorithms are 1.95% for Toeplitz-connected, 2.10% for the binary-connected, and 2.20% for the fully connected network. In general, the modified Broyden-Fletcher-Goldfarb-Shanno (BFGS) methods, the three variants of LM algorithm, and the new hybrid/LS method perform consistently well, achieving error rates of less than 3% averaged across all three architectures.