• The Transactions of the Korea Information Processing Society
• /
• v.2 no.2
• /
• pp.215-228
• /
• 1995

Despite their importance in language, there have been relatively few computational studies in understanding words. This paper describes how neural networks can learn to perceive and produce words. Most traditional linguistic theories presuppose abstract underlying representations (UR) and a set of explicit rules to obtain the surface realization. There are, however, a number of questions that can be raised regarding this approach: (1) assumption of URs, (2) formation of rules, and (3) interaction of rules. In this paper, it is hypothesized that rules would emerge as the generalizations the network abstracts in the process of learning to associate forms with meanings of the words. Employing a simple recurrent network, a series of simulations on different types of morphophonemic processes was run. The results of the simulations show that this network is capable of learning to perceive whether words are in basic from or in inflected form, given only forms, and to produce words in the right form, given arbitrary meanings, this eliminating the need for presupposing abstract URs and rules.

• Wind and Structures
• /
• v.36 no.3
• /
• pp.149-160
• /
• 2023

Lightness and flexibility of membrane roofs make them very sensitive to any external load. One of the most important parameters that controls their behavior, especially under wind load is the sag/span ratio of edge cables. Based on the value of the pretension force in the edge cables and the horizontal projection of the actual area covered by the membrane, an optimized design range of cable sag/span ratios has been determined through carrying on several membrane form-finding analyses. Fully coupled fluid structure dynamic analyses of these membrane roofs are performed under wind load with several conditions using the CFD method. Through investigating the numerical results of these analyses, the behavior of membrane roofs with cables sag/span ratios selected from the previously determined optimized design range has been evaluated.

• International Journal of Fluid Machinery and Systems
• /
• v.2 no.4
• /
• pp.303-314
• /
• 2009

The present paper shows the results of numerical and experimental modal analyses of Francis runners, which were executed in air and in still water. In its first part this paper is focused on the numerical prediction of the model parameters by means of FEM and the validation of the FEM method. Influences of different geometries on modal parameters and frequency reduction ratio (FRR), which is the ratio of the natural frequencies in water and the corresponding natural frequencies in air, are investigated for two different runners, one prototype and one model runner. The results of the analyses indicate very good agreement between experiment and simulation. Particularly the frequency reduction ratios derived from simulation are found to agree very well with the values derived from experiment. In order to identify sensitivity of the structural properties several parameters such as material properties, different model scale and different hub geometries are numerically investigated. In its second part, a harmonic response analysis is shown for a Francis runner by applying the time dependent pressure distribution resulting from an unsteady CFD simulation to the mechanical structure. Thus, the data gained by modern CFD simulation are being fully utilized for the structural design based on life time analysis. With this new approach a more precise prediction of turbine loading and its effect on turbine life cycle is possible allowing better turbine designs to be developed.