Finite element analysis of a shear punch test

The authors present new experimental data for shear punch tests and tensile tests using four different materials. The correlation between shear punch and tensile yield stresses must be established empirically, and the procedure for doing this is discussed. The elastic-plastic deformation in a shear punch test develops gradually with increasing punch displacement and is often assumed to be simple shear. This is examined using finite element analysis to simulate the development of plasticity during the early stages of punch displacement. The simulation results are used to rationalize the empirically established yield stress correlations.     

A knowledge-based system for the recycling of non-hazardous industrial residuals in civil engineering applications

Abstract: It is estimated that 4.6 billion tons of non‐hazardous solid waste materials are produced annually in the USA. The potential reuse for a portion of the materials in the construction of highways and roads suggests that valuable benefits in terms of economic and environmental gains are possible. This paper describes the development of a prototype computer‐assisted tool or expert system to help manufacturers assess and analyze their industrial residuals as potential road construction material. This represents an expansion in the application of intelligent systems to domains where a few, hard‐to‐find technical reports have represented the main source of expertise available to practitioners for years. The system, developed through the use of an object‐oriented software shell, Level5 Object, was designed in a user‐friendly Windows environment which allows users with little or no computer training to effectively evaluate material residuals.     

Improved global robust asymptotic stability criteria for delayed cellular neural networks.

This paper considers the problem of global robust stability analysis of delayed cellular neural networks (DCNNs) with norm-bounded parameter uncertainties. In terms of a linear matrix inequality, a new sufficient condition ensuring a nominal DCNN to have a unique equilibrium point which is globally asymptotically stable is proposed. This condition is shown to be a generalization and improvement over some previous criteria. Based on the stability result, a robust stability condition is developed, which contains an existing robust stability result as a special case. An example is provided to demonstrate the reduced conservativeness of the proposed results.     

Designing stable MIMO fuzzy controllers.

This paper presents a systematic procedure for constructing a multi-input multi-output fuzzy controller that guarantees identical performance to an existing stabilizing linear controller. An algorithm is devised that generates a fuzzy controller which is functionally identical to a given time-invariant or time-varying finite-dimensional linear controller. The benefit of this transformation is that it provides an automated technique for the initial fuzzy controller setup while vital knowledge-based attributes are integrated afterwards. An important result of this work is that once a linguistic mapping into the fuzzy domain has been performed, one can see in linguistic terms how the linear controller operates. The effectiveness of this approach is demonstrated with a model for a flexible robot that exhibits nonminimum phase characteristics. An extension is outlined to use this deterministic approach for the case of a general dynamic control-law, and several applications to nonlinear control problems are discussed.     

A Reinforcement Learning Algorithm in Cooperative Multi-Robot Domains

Reinforcement learning has been widely applied to solve a diverse set of learning tasks, from board games to robot behaviours. In some of them, results have been very successful, but some tasks present several characteristics that make the application of reinforcement learning harder to define. One of these areas is multi-robot learning, which has two important problems. The first is credit assignment, or how to define the reinforcement signal to each robot belonging to a cooperative team depending on the results achieved by the whole team. The second one is working with large domains, where the amount of data can be large and different in each moment of a learning step. This paper studies both issues in a multi-robot environment, showing that introducing domain knowledge and machine learning algorithms can be combined to achieve successful cooperative behaviours.     

Satellite-derived parameters for biological modelling in coastal waters: Illustration over the eastern continental shelf of the Bay of Biscay

In biological modelling of the coastal phytoplankton dynamics, the light attenuation coefficient is often expressed as a function of the concentrations of chlorophyll and mineral suspended particulate matter (SPM). In order to estimate the relationship between these parameters over the continental shelf of the northern Bay of Biscay, a set of in situ data has been gathered for the period 1998-2003 when SeaWiFS imagery is available. These data comprise surface measurements of the concentrations of total SPM, chlorophyll, and irradiance profiles from which is derived the attenuation coefficient of the photosynthetically available radiation, K_PAR. The performance of the IFREMER look-up table used to retrieve the chlorophyll concentration from the SeaWiFS radiance is evaluated on this new set of data. The quality of the estimated chlorophyll concentration is assessed from a comparison of the variograms of the in situ and satellite-derived chlorophyll concentrations. Once the chlorophyll concentration is determined, the non living SPM, which is defined as the SPM not related to the dead or alive endogenous phytoplankton, is estimated from the radiance at 555 nm by inverting a semi-analytic model. This method provides realistic estimations of concentrations of chlorophyll and SPM over the continental shelf all over the year. Finally, a relationship, based on non living SPM and chlorophyll, is proposed to estimate K_PAR on the continental shelf of the Bay of Biscay. The same formula is applied to non living SPM and chlorophyll concentrations, observed in situ or derived from SeaWiFS radiance.     

On the semantics of noun compounds

This paper provides new insights on the semantic characteristics of two and three noun compounds. An analysis is performed using two sets of semantic classification categories: a list of 8 prepositional paraphrases previously proposed by Lauer [Designing statistical language learners: experiments on noun compounds, Ph.D. Thesis, Macquarie University, Australia] and a new set of 35 semantic relations introduced by us. We show the distribution of these semantic categories on a corpus of noun compounds and present several models for the bracketing and the semantic classification of noun compounds. The results are compared against state-of-the-art models reported in the literature.     

Elimination of artifacts due to occlusion and discretization problems in image space blurring techniques

Traditional computer graphics methods render images that appear sharp at all depths. Adding blur can add realism to a scene, provide a sense of scale, and draw a viewer’s attention to a particular region of a scene. Our image-based blur algorithm needs to distinguish whether a portion of an image is either from a single object or is part of more than one object. This motivates two approaches to identify objects after an image has been rendered. We illustrate how these techniques can be used in conjunction with our image space method to add blur to a scene.     

Forecasting Economic Data with Neural Networks

Studies in recent years have attempted to forecast macroeconomic phenomena with neural networks reporting mixed results. This work represents an investigation of this problem using U.S. Real Gross Domestic Production and Industrial Production as case studies. This work is based on a coefficient of determination which accurately measures the ability of linear or nonlinear models to forecast economic data. The significance of our work is twofold: (1) It confirms recent work that neural networks significantly outperform linear regression due to nonlinearities inherent in the data sets, and (2) it provides a systematic approach that guarantees to find the maximum correlation between input(s) and output of interest. JEL Classification: C45, C52, C53, C67,