The neural network approach has been used for foF2 long-term prediction one hour ageas. The feedfoward multilayer structure has been applied. The neural network was trained on the basis of the data from Uppsala station. The results of numerical experiments performed by using real of foF2 show a satisfactory agreement with measurements.
2
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Assembler Encoding represents neural network in the form of a simple program called Assembler Encoding Program. The task of the program is to create the so-called Network Definition Matrix which maintains the whole information necessary to construct the network. To generate Assembler Encoding Programs and in consequence neural networks evolutionary techniques are used. In order to use evolutionary techniques to construct Assembler Encoding Programs it is necessary to encode them in the form of chromosomes. The simplest solution is to place the whole information necessary to construct the program into one chromosome. The paper suggests another approach. Methods proposed in the paper are an adaptation of Symbiotic Adaptive Neuro-evolution. To test the methods proposed they were used to solve a few simple optimization problems.
3
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
The primary objective of this paper is to design a method of detecting road edges without using complex algorithms to identify and analyze images. Instead, neural networks are used, which allows to enhance and facilitate this process. The paper describes a program that recognizes the road in a picture with the use of a neural network trained on 500 samples. The samples contain original photos and images with a selected road. In the course of the research two solutions arose. The first solution is to use a single Perceptron to recognize the road. The second solution is to classify the photos using a Kohonen network and establish a separate network for each class.
4
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Artificial intelligence methods for MT data processing are proposed. Distortions having a complex structure created by external artificial sources such as, for example, passing train were investigate. In the first part of this paper the time intervals with such type of distortions were found by using a special neuronal system. Next for time intervals found in the first stage the measure curve fragment is removed and then it is replied by the fragment created by a trained perceptron. The experiment showed that used method are effective.
Due to load uncertainties of cranes, it is necessary to find exact kinematic parameters of crane mechanisms. This research is concerned with application of neural network to the force analysis of a crane mechanism. The type of network investigated is a Recurrent Hybrid Neural Network (RHNN). The crane mechanism is considered as a double-rocker four-bar mechanism. Desired kinematic parameteres of the crane is found by a software dealing with simulation and analysis of mechanisms. The RHNN is employed in four parameters prediction schemes: displacement, velocity, acceleration and joints forces. The results obtained have supported the theory that the proposed RHNN is able to predict different types of crane system.
The paper is designed as a summary of the relatively wide study on usage of Evolutionary Algorithms (EAs) as an automatic tool for Neural Networks (NNs) design. The study was conducted for many years and partial results have been presented. Special attention is put to such aspects of NNs evolution as chromosome coding, defining fitness function, and different evolutionary approaches. The studied approaches are shortly described and only general results, summing up all experiments are presented. Summary contains some recommendations formulated on the basis of experiments. Future works are also mentioned.
This paper is concerned with an overview of artificial neural networks from the point of view of the possibility of using them to process information as well as an overview of new constructions of artificial neural networks that can be used to process knowledge about electrical energy market. In particular, attention has been drawn to self-organizing neural networks and self-optimizing neural networks, which show the highest sensitivity to neural network design for examining the regularities in the operation and development of great systems of this order such as the system of electrical energy market.
8
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
The paper deals with an application of neural networks for computation of fundamental natural periods of buildings with load-bearing walls. The identification problem is formulated as a relation between structural and soil parameters and the fundamental period of building. The patterns are based on long-term tests performed bn actual structures. Various splitting up of the set of patterns into training and testing sets are considered in the analysis. The carried out analysis leads to conclusion that, even in "the worst" case of randomly selected testing patterns, the natural periods of vibrations of buildings are obtained with accuracy quite satisfactory for engineering practice.
9
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
This paper proposes a neural network model using genetic algorithm for a model for the prediction of the damage condition of existing light structures founded in expansive soils in Victoria, Australia. It also accounts for both individual effects and interactive effects of the damage factors influencing the deterioration of light structures. A Neural Network Model was chosen because it can deal with 'noisy' data while a Genetic Algorithm was chosen because it does not get 'trapped' in local optimum like other gradient descent methods. The results obtained were promising and indicate that a Neural Networlc Model trained using a Genetic Algorithm has the ability to develop an interactive relationship and a Predicted Damage Conditions Model.
In this article an application of neural networks to the reconstruction of unknown physical quantities in particle physics is presented. As an example the mass reconstruction of the hypothetical Higgs boson in the typical high energy physics experiment is used. Monte Carlo events are used to determine the probability distributions of observables (energies of two jets and the angle between them) for various Higgs boson mass, which are later fitted using a Neural Network. These distributions are used to determine the mass probability distribution of the measured particle. The mass is reconstructed without knowing the functional dependence between the observables and the measured quantity. The miscalibration of the measured quantities is automatically corrected in this method.
PL
W artykule zaprezentowane jest zastosowanie sieci neuronowych do rekonstrukcji nieznanych wielkości w fizyce cząstek elementarnych. Jako przykład użyta jest rekonstrukcja masy hipotetycznego bozonu Higgsa oparta na symulowanych danych. Dane te zostały użyte do wyznaczenia rozkładów prawdopodobieństwa mierzonych wielkości (energie dwóch dżetów oraz kąt pomiędzy nimi) dla różnych mas cząstki Higgsa. Rozkłady te zostały następnie sparametryzowane za pomocą sieci neuronowych oraz wyznaczenia rozkładu prawdopodobieństwa masy mierzonej cząstki. Masa jest wyznaczona bez użycia zależności funkcyjnej pomiędzy mierzonymi wielkościami a rekonstruowaną masą. Kalibracja wielkości pomiarowych jest automatycznie korygowana poprzez rozkłady prawdopodobieństwa.
11
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
The past few years have witnessed a dynamic growth variety of neural network applications. Range of these applications is very wide especially in industrial process control. As in nature, the neural network is determined by the connections between the elements, and we can train its to perform the particular function by adjusting special values (weights between elements) [1,4,7]. This paper presents DC motor model controlled by neural network Proportional-Derivate controller in comparison with classic PD controller. The research concern different network architectures and training functions. The models are presented and the experimental results signals are shown using graphical charts.
12
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Zastosowanie sieci neuronowych do prognozowania zachowania rynków finanso-wych jest jednym z ich najpopularniejszych zastosowań. W artykule przedstawiono re-zultaty badania możliwości przewidywania wskaźnika WIG oraz wyznaczenia firm o największym ważonym wpływie na WIG. Przedstawiono również próby wyznaczenia trendów czasowych na Giełdzie Warszawskiej w okresie 1996-1997.
EN
Use of neural networks to financial markets forecasting is one of the most popular applications. Results of the attempts for forecasting WIG indices, and estimating the company's with the biggest weighted impact on WIG are presented in this article. Attempts for estimating time trends on Warsaw Stock Exchange in 1996-1997 are also highlighted in that article.
The objective of this research was to use neural network approach for segmentation problem of agricultural landed-fields in remote sensing data. A neural network clusterization algorithm for segmentation of the color images of crop field infected by diseases that change usual color of agricultural plants is proposed. It can be applied for cartography of fields infected by plant diseases to reduce the use of plant protection products.
14
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Fabric defect detection and classifcation plays a very important role in the automatic detection process for fabrics. This paper refers to the seven commonly seen defects of lycra spandex: laddering, end-out, hole, oil spot, dye stain, snag, and crease mark. First of all, the gray level co-occurrence matrix was used to collect the features of the fabric image texture, and then the back-propagation neural network was used to establish faw classifcations of the fabric. In addition, by using the Taguchi method combined with BPNN, the BPNN drawback was improved upon, which requires overly time consuming trial-and-error to fnd the learning parameters, and could therefore converge even faster with an even smaller convergence error and better recognition rate. The experimental results proved that the fnal root-mean-square error convergence of the Taguchi-based BPNN was 0.000104, and that the recognition rate can reach 97.14%.
15
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Austempered ductile iron is an excellent material and it possesses attractive properties as high strength, ductility and toughness. [n this paper the neural technique was applied to the analysis of the ultimate tensile strength, the yield strength and elongation of austempered ductile iron (ADI). Estimation of the mechanical properties of austempered ductile cast iron (ADI) was curried as a function of chemical composition and heat treatment conditions (austenitizing temperature, austenitizing time, austempering temperature and austempering time). A 'committee' model was used to increase the accuracy of the predictions. The model was validated by comparison its predictions with data of tensile tests experiments on austempered samples of ductile cast iron. The model successfully reproduces experimentally determined ultimate tensile strength and elongation and it can be exploited in the predictions of tensile properties in the design of chemical composition of cast irons and their heat treatments.
16
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
W pracy zaprezentowano nowy sposób szacowania wskaźników ruchu w lokomocji dwunożnej człowieka, wykorzystywanych w tzw. metodzie wskaźnikowej. Otrzymane wyniki świadczą o przydatności sztucznych sieci neuronowych do analizy i oceny energetycznej chodu człowieka. Całość zaprezentowano na tle obecnie stosowanych urządzeń do analizy i metod oceny lądowej lokomocji człowieka w polu grawitacyjnym Ziemi.
EN
The new method of assessment human gait indicators has been presented. The results show the possibility of artificial neural networks in analysis and human gait assessment. A review of actually using measurement systems and methods of human gait assessment have been also presented.
One of the major limitations of productivity and quality in metal cutting is machining accuracy of machine tools. The machining accuracy is affected by geometric errors, thermally induced errors, and deterioration of the machine tools. Geometric and thermal errors of machine tools. Geometric and thermal errors of machine tools should be measured and compensated to manufacture high quality products. In metal cutting, the machining accuracy is more affected by thermal errors than by geometric errors. This paper models the thermal errors for error analysis and develops an on the machine measurement system by which the volumetric errors are measured and compensated. The thermal error is modeled means of angularity errors of column and thermal drift error pf the spindle unit which are measured by the touch probe unit with a star type styluses and a designed spherical ball artifact (SBA). Experiments show that the developed system provides a high measuring accuracy, with repeatability of +- 2um in X,Y and Z directions. A neural network model is used for predicting thermal errors occurring while in process. The thermal errors are compensated by means of a custom macro of the machine tool controller. With the compensation effort, the machining accuracy of machine tools is improved. It is believed that the developed measurement system can be also applied to the machine tools with CNC controller. In addition, machining accuracy and product quality can be also improved by using the developed measurement when the spherical ball artifact is mounted on a modular fixture.
PL
Jednym z głównych ograniczeń wydajności i jakości w skrawaniu metali jest dokładność skrawania. Na dokładność skrawania mają wpływ błędy geometryczne, błędy spowodowane efektami cieplnymi oraz pogarszanie się jakości obrabiarek. Błędy geometryczne i cieplne obrabiarek należy mierzyć i kompensować, aby wytwarzać produkty wysokiej jakości. W przypadku skrawania metali na dokładność skrawania wpływają bardziej błędy cieplne niż błędy geometryczne. Autorzy referatu zamodelowali błędy cieplne dla celów analizy błędów i opracowali układ pomiarowy na obrabiarce, za pomocą którego błędy objętościowe są mierzone i kompensowane. Błąd cieplny jest modelowany za pomocą błędów nachylenia kolumny i błędu dryfu cieplnego zespołu wrzeciona, mierzonych sondą dotykową z palcami prowadzącymi typu gwiazda i sferycznym artefaktem kulkowym (spherical ball artifact). Badania wykazały, że opracowany układ zapewnia wysoką dokładność pomiaru z powierzchnią rzędu +-2um w kierunkach X,Y,Z. Model sieci neuronowej zastosowano do przewidywania błędów cieplnych występujących w trakcie procesu. Błędy cieplne kompensowane są za pomocą projektowanego na zamówienie makra sterownika obrabiarki. Dzięki tej kompensacji dokładność skrawania obrabiarek ulega poprawie. Uważa się, że opracowany układ pomiarowy można zastosować także do obrabiarek ze skomputeryzowanym sterowaniem numerycznym. Ponadto dokładność obróbki i jakość wyrobu można poprawić przez zastosowanie opracowanego układu pomiarowego, gdy sferyczny artefakt kulkowy jest zamocowany na modularnym uchwycie specjalnym.
19
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Analysing arid predicting the detailed three dimensional conformation of protein structures is a critical and important task within structural bioinformatics with impact on other fields, e.g.. drug design and delivery, sensing technologies, etc. Unfortunately, it is hard to identify one methodology that will give the best prediction of the three-dimensional structure for any sequence. That is, some predictors are best suited for some sequences and not for others. In trying to address this drawback of current prediction algorithms the research community introduced the concept of protein prediction metaservers. In this paper we propose a new metaserver method called 3D-Judge that uses an artificial neural network (ANN) to select the best model from among models produced by individual servers. The fundamental innovation we introduce is that the AXN is not only used to decide which models and servers to use as good predictions but, crucially, it is also used to analyse and "remember" the past performances of the servers it has access to. Thus, our method acts as both a kind of majority voting algorithm, by selecting models arising from different servers based on their mutual similarity, and also a reinforced learning method that takes cues from historical data of previously solved structures. We train and evaluate our metaserver based on previous GASP results and we compare SD-Judge with a popular and effective metaserver, namely. 3D-Jury. The obtained results indicate that 3D-Judge is competitive with 3D-Jury, outperforming it on many cases. We also present a discussion on future extensions to 3D-Judge.
20
Dostęp do pełnego tekstu na zewnętrznej witrynie WWW
Artificial Neural Network (ANN) has been widely utilized in the past ten years in civil engineering applications for the simulation and prediction of the different physical phenomen and has proven its capabilities in the different fields. The existence of hydraulic structures in a branched open channel system urges the need for considering the gradually varied flow criterion in evaluating the different hydra ulic characteristics in this type of open channel system. Computations of hydraulic characteristics such as flow rates and water surface profiles in branched open channel system with hydraulic structures require tremendous numerical effort especially when the flow cannot be assumed uniform. The present study aims towards introducing the use of ANN technique to model and predict the impact of changing water structures' locations on the hydraulic performance of branched open channel system. Specifically the current paper investigates α branched open channel system that consists of main channel supplies water to two branch channels with water structures such as clear over fall weirs and sluice gates. The results of this study showed that ANN technique was capable, with small computational effort and high accuracy, of predicting the impact of changing the locations of two types of water structures on the hydraulic performance of branched open channel system.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.