Using recurrent neural networks for adaptive communication channelequalization

Nonlinear adaptive filters based on a variety of neural network models have been used successfully for system identification and noise-cancellation in a wide class of applications. An important problem in data communications is that of channel equalization, i.e., the removal of interferences introduced by linear or nonlinear message corrupting mechanisms, so that the originally transmitted symbols can be recovered correctly at the receiver. In this paper we introduce an adaptive recurrent neural network (RNN) based equalizer whose small size and high performance makes it suitable for high-speed channel equalization. We propose RNN based structures for both trained adaptation and blind equalization, and we evaluate their performance via extensive simulations for a variety of signal modulations and communication channel models. It is shown that the RNN equalizers have comparable performance with traditional linear filter based equalizers when the channel interferences are relatively mild, and that they outperform them by several orders of magnitude when either the channel's transfer function has spectral nulls or severe nonlinear distortion is present. In addition, the small-size RNN equalizers, being essentially generalized IIR filters, are shown to outperform multilayer perceptron equalizers of larger computational complexity in linear and nonlinear channel equalization cases.     

Extensible plastic collapse of thin-wall frusta as energy absorbers

The crumpling of thin-walled frusta, under axial compression, in the ‘concertina’ mode is studied. The energy expended in bending at the plastic hinges and in stretching the metal between the hinges is minimized for the total decrease in height due to collapse. The thinning of the cross-section due to stretching is neglected. A theoretical model has been developed and numerical results are obtained that show the effect of slenderness, , and the semi-apical angle of the frusta. Good qualitative agreement in trends is exhibited when comparison with available experimental results is made.     

Bibliometric analysis of the evolution of biochar research trends and scientific production

Clean Technologies and Environmental Policy - Biochar has caught great attention over the last decade, yielding a large number of publications in a broad range of disciplines. This scientometric...     

Analytical and experimental approach to damage and residual strength of fibreglass composite automotive frame rails during manufacturing

The use of composite materials in the construction of various structural components of the car body has gained an increasing interest in the automotive industry. In relation to this, the crashworthy ability of an automotive frame rail of hour glass cross-section, made of a glass fiber/ vinylester composite and designed for use in the apron construction of the car, when subjected to axial collapse and bending, has been extensively investigated in this paper.

An important problem arose, dealing with the delamination observed at the corners and curved parts of the hourglass section which may be attributed to the manufacturing processing, resulting in severe limitations in the load-carrying capacity of this structural component. The effect of the residual stress and strain fields induced during fabrication was examined both theoretically and experimentally and possible failure mechanisms observed were investigated and discussed.     

一种新型的夹芯结构

金属（铝壳,蜂房或金属泡沫核）或聚合物（复合面层,聚合物泡沫核）夹芯结构被认为是承受弯曲荷载的优化设计结构。本文在发挥金属性材料和聚合物材料优势的基础上,研制了一种新的混杂夹芯结构。这种新理念的结构中金属片用在外表面以加强刚度,轻质核与外壳粘结成整体。此外,还将复合层或木材层作为中间层以提高冲击阻力。制造这种新结构潜在的方法是基于真空压缩。对该结构的研究包含基于有限元分析的几种面层的理论构造,改进的简化方程,对典型案例的试验。     

Unsupervised statistical neural networks for model-based objectrecognition

Statistical neural networks executing soft-decision algorithms have been shown to be very effective in many classification problems. A neural network architecture is developed here that can perform unsupervised joint segmentation and labeling of objects in images. We propose the semi-parametric hierarchical mixture density (HMD) model as a tool for capturing the diversity of real world images and pose the object recognition problem as a maximum likelihood (ML) estimation of the HMD parameters. We apply the expectation-maximization (EM) algorithm for this purpose and utilize ideas and techniques from statistical physics to cast the problem as the minimization of a free energy function. We then proceed to regularize the solution thus obtained by adding smoothing terms to the objective function. The resulting recursive scheme for estimating the posterior probabilities of an object's presence in an image corresponds to an unsupervised feedback neural network architecture. We present here the results of experiments involving recognition of traffic signs in natural scenes using this technique     

Finite element simulation of the axial collapse of metallic thin-walled tubes with octagonal cross-section

The present paper deals with the implementation of the explicit FE Code LS-DYNA to simulate the crash behaviour and energy absorption characteristics of steel thin-walled tubes of octagonal cross-section subjected to axial loading. The collapse procedure is successfully simulated and the obtained numerical results are compared with actual experimental data from small-scale models and useful concluding remarks pertaining to the design requirements of the crushing process are drawn.     

Distributed Memory Parallel Architecture Based on Modular Linear Arrays for 2-D Separable Transforms Computation

A framework for mapping systematically 2-dimensional (2-D) separable transforms into a parallel architecture consisting of fully pipelined linear array stages is presented. The resulting model architecture is characterized by its generality, high degree of modularity, high throughput, and the exclusive use of distributed memory and control. There is no central shared memory block to facilitate the transposition of intermediate results, as it is commonly the case in row-column image processing architectures. Avoiding shared central memory has positive implications for speed, area, power dissipation and scalability of the architecture. The architecture presented here may be used to realize any separable 2-D transform by only changing the coefficients stored in the processing elements. Pipelined linear arrays for computing the 2-D Discrete Fourier Transform and 2-D separable convolution are presented as examples and their performance is evaluated.     

A new hybrid concept for sandwich structures

Sandwich structures are considered as optimal designs for carrying bending loads and can be either metal (aluminium faces and honeycomb or metal foam cores) or polymer structures (composite faces with polymer foam cores). In this paper, a new hybrid sandwich structure has been developed by combining most of the advantages of metallic and polymeric materials while avoiding some of their main disadvantages. For this new concept metal sheets are used at the outer surfaces to maximize rigidity while introducing in between lightweight cores adhesively bonded to keep the whole structure together. Furthermore, composite or wood layers may be used as intermediate layers to improve impact resistance. Potential methods for the manufacturing of this new structure are based on compression under vacuum. The results include the study of several panel configurations theoretically based on Finite element analysis and on the modified simplified equations and experimental results in the most representative cases of the study.     

Surface roughness prediction in turning of femoral head

This paper is dealing with the development of a surface roughness model for turning of femoral heads from AISI 316L stainless steel. The model is developed in terms of cutting speed, feed rate and depth of cut, using response surface methodology. Machining tests were carried out with TiN–Al₂O₃–TiC-coated carbide cutting tools under various conditions. First-order and second-order models predicting equations for surface roughness have been established by using the experimental results. The established equation shows that the depth of cut was the main influencing factor on the surface roughness. It increased with increasing the depth of cut and feed rate, respectively, but it decreased with increasing the cutting speed. In addition, analysis of variance for the second-order model shows that the interaction terms and the square terms are statistically insignificant. The predicted surface roughness of the samples was found close to the experimentally obtained results within a 95% confident interval.