Wyniki wyszukiwania - BazTech

1

Indirect encoding in neuroevolutionary ship handling

Łącki M.

TransNav : International Journal on Marine Navigation and Safety of Sea Transportation

|

2018

|

Vol. 12, no. 1

71--76

EN

In this paper the author compares the efficiency of two encoding schemes for artificial intelligence methods used in the neuroevolutionary ship maneuvering system. This may be also be seen as the ship handling system that simulates a learning process of a group of artificial helmsmen - autonomous control units, created with an artificial neural network. The helmsman observes input signals derived form an enfironment and calculates the values of required parameters of the vessel maneuvering in confined waters. In neuroevolution such units are treated as individuals in population of artificial neural networks, which through environmental sensing and evolutionary algorithms learn to perform given task efficiently. The main task of this project is to evolve a population of helmsmen with indirect encoding and compare results of simulation with direct encoding method.

2

Learning Search Algorithms: An Educational View

Janota M., Šimák V., Hrbček J.

TransNav : International Journal on Marine Navigation and Safety of Sea Transportation

|

2014

|

Vol. 8 no. 4

565--570

EN

Artificial intelligence methods find their practical usage in many applications including maritime industry. The paper concentrates on the methods of uninformed and informed search, potentially usable in solving of complex problems based on the state space representation. The problem of introducing the search algorithms to newcomers has its technical and psychological dimensions. The authors show how it is possible to cope with both of them through design and use of specialized authoring systems. A typical example of searching a path through the maze is used to demonstrate how to test, observe and compare properties of various search strategies. Performance of search methods is evaluated based on the common criteria.

3

Metody pomiaru zużycia energii elektrycznej przez odbiorniki oraz ich identyfikacji w energooszczędnym budynku inteligentnym

Bilski P.

Elektro Info

|

2014

|

nr 6

77--81

PL

W artykule przedstawiono metodykę pomiarów oraz identyfikacji urządzeń elektrycznych w domu inteligentnym na podstawie pomiaru mocy wydzielającej się w obciążeniu. Ze względu na obecne trendy zmierzające do minimalizacji zużycia energii przez gospodarstwa domowe na całym świecie jest to jedno z priorytetowych za-dali stojących przed społeczeństwami rozwiniętymi. Z tego powodu przedstawiono ogólny schemat systemu poboru energii elektrycznej w budynku wraz z klasyfikacją urządzeń w nim wykorzystywanych. W pierwszej kolejności omówiono techniki pomiarowe oraz parametry uzyskiwane za pomocą czujników i mierników, będące podstawą do identyfikacji poszczególnych odbiorników. Następnie przedstawiono klasyfikację metod służących do wykrywania działania poszczególnych rodzajów urządzeń. Ponieważ do tego celu stosowane są głównie metody sztucznej inteligencji, w artykule skupiono się głównie na nich. Na końcu przedstawione zostały wnioski i uwagi na temat potencjalnego rozwoju metodyki.

EN

The paper presents the methodology of measurement and identification of electrical appliances in the smart house based on the power produced in the load. Because of the current trends leading to minimize the energy consumption in households across the whole world, this is one of the priorities in the developed countries. Firstly, the general scheme of the energy collection system and classiciation of appliances are presented. Then, measurement techniques and symptoms acquired by sensors and monitoring devices are discussed. The latter are used to identify subsequent groups of appliances. The taxonomy of methods used to identify the appliances based on the set of symptoms is introduced. They are mainly artificial intelligence approaches, being the main focus in the paper. Finally, conclusions and future prospects about the potential implementation of the methodology are presented.

4

Numerical simulation of the alloying elements effect on steels’ properties

Sitek W., Trzaska J.

Journal of Achievements in Materials and Manufacturing Engineering

|

2011

|

Vol. 45, nr 1

71--78

EN

Purpose: The goal of the research carried out was evaluation of alloying elements effect on high-speed steels hardness and fracture toughness and austenite transformations during continuous cooling of structural steels. Design/methodology/approach: Multi-layer feedforward neural networks with learning rule based on the error backpropagation algorithm were employed for modelling the steels properties. Then the neural networks worked out were employed for the computer simulation of the effect of particular alloying elements on the steels’ properties. Findings: Obtained results show that neural network are useful in evaluation of synergic effect of alloying elements on selected materials properties when classical investigations’ results do not provide evaluation of the effect of two or more alloying elements. Practical implications: Numerical simulation presented in the work, based on using the adequate material models may feature an alternative for classical investigations on effect of alloying elements on steels’ properties. Originality/value: The use of the neural networks as an tool for evaluation of the chemical composition effect on steels’ properties.

5

Methodology of high-speed steels design using the artificial intelligence tools

Sitek W.

Journal of Achievements in Materials and Manufacturing Engineering

|

2010

|

Vol. 39, nr 2

115-160

EN

Purpose: The main goal of the research carried out was developing the design methodology for the new high-speed steels with the required properties, including hardness and fracture toughness, as the main properties guaranteeing the high durability and quality of tools made from them. It was decided that hardness and fracture toughness KIc are the criteria used during the high-speed steels design. Design/methodology/approach: In case of hardness, the statistical and neural network models were developed making computation possible of the high-speed steel hardness based solely on the steel chemical composition and its heat treatment parameters, i.e., austenitizing- and tempering temperatures. In this case results of own work on the effect of alloy elements on the secondary hardness effect were used, as well as data contained in catalogues and pertinent standards regarding the high-speed steels. In the second case - high-speed steels fracture toughness, the neural network model was developed, making it possible to compute the KIc factor based on the steel chemical composition and its heat treatment parameters. The developed material models were used for designing the chemical compositions if the new high-speed steel, demonstrating the desired properties, i.e., hardness and fracture toughness. Methodology was developed to this end, employing the evolutionary algorithms, multicriteria optimisation of the high-speed steels chemical composition. Findings: Results of the research carried out confirmed the assumption that using the data from catalogues and from standards is possible, which - would supplement the set of data indispensable to develop the assumed model - improving in this way its adequacy and versatility. Practical implications: Solutions presented in the work, based on using the adequate material models may feature an interesting alternative in designing of the new materials with the required properties. The practical aspect has to be noted, resulting form the developed models, which may successfully replace the above mentioned technological investigations, consisting in one time selection of the chemical composition and heat treatment parameters and experimental verification of the newly developed materials to check of its properties meet the requirements. Originality/value: The presented approach to new materials design, being the new materials design philosophy, assumes the maximum possible limitation of carrying out the indispensable experiments, to take advantage of the existing experimental knowledge resources in the form of databases and most effective computer science tools, including neural networks and evolutionary algorithms. It should be indicated that the materials science knowledge, pertaining oftentimes to the multi-aspect classic problems and described, or - rather - saved in the existing, broadly speaking, databases, features the invaluable source of information which may be used for discovery of the unknown so far relationships describing the material structure - properties relations.

6

Artificial intelligence-based control system for the analysis of metal casting properties

Mares E., Sokolowski J. H.

Journal of Achievements in Materials and Manufacturing Engineering

|

2010

|

Vol. 40, nr 2

149-154

EN

Purpose: The metal casting process requires testing equipment that along with customized computer software properly supports the analysis of casting component characteristic properties. Due to the fact that this evaluation process involves the control of complex and multi-variable melting, casting and solidification factors, it is necessary to develop dedicated software. Design/methodology/approach: The integration of Statistical Process Control methods and Artificial Intelligence techniques (Case-Based Reasoning) into Thermal Analysis Data Acquisition Software (NI LabView) was developed to analyze casting component properties. The thermal data was tested in terms of accuracy, reliability and timeliness in order to secure metal casting process effectiveness. Findings: Quantitative values were defined as “Low”, “Medium” and “High” to assess the level of improvement in the metal casting analysis by means of the Artificial Intelligence-Based Control System (AIBCS). The traditional process was used as a reference to measure such improvement. As a result, the accuracy, reliability and timeliness were significantly increased to the “High” level. Research limitations/implications: Presently, the AIBCS predicts a limited number of casting properties. Due to its flexible design more properties could be added. Practical implications: The AIBCS has been successfully used at the Ford/Nemak Windsor Aluminum Plant (WAP) to analyze Al casting properties of the engine blocks. Originality/value: The metal casting research community has immensely benefited from these developed information technologies that support the metal casting process.

7

Application of artificial neural networks in modelling of quenched and tempered structural steels mechanical properties

Dobrzański L. A., Honysz R.

Journal of Achievements in Materials and Manufacturing Engineering

|

2010

|

Vol. 40, nr 1

50-57

EN

Purpose: This paper presents the application of artificial neural networks for mechanical properties prediction of structuralal steels after quenching and tempering processes. Design/methodology/approach: On the basis of input parameters, which are chemical composition, parameters of mechanical and heat treatment and dimensions of elements, steels’ mechanical properties : yield stress, tensile strength stress, elongation, area reduction, impact strength and hardness are predicted. Findings: Results obtained in the given ranges of input parameters indicates on very good ability of artificial neural networks to values prediction of described mechanical properties for steels after quenching and tempering processes. The uniform distribution of descriptive vectors in all, training, validation and testing sets, indicates on good ability of the networks to results generalisation. Practical implications: Artificial neural networks, created during modelling, allows easy prediction of steels properties and allows the better selection of both chemical composition and the processing parameters of investigated materials. It’s possible to obtain steels, which are qualitatively better, cheaper and more optimised under customers needs. Originality/value: The prediction possibility of the material mechanical properties is valuable for manufacturers and constructors. It allows the preservation of customers quality requirements and brings also measurable financial advantages

8

Application of artificial neural networks for modelling correlations in age hardenable aluminium alloys

Musharavati F., Hamouda A. S. M.

Journal of Achievements in Materials and Manufacturing Engineering

|

2010

|

Vol. 41, nr 1-2

140-146

EN

Purpose: This paper discusses some of the preliminary results of an ongoing research on the applications of artificial neural networks (ANNs) in modelling, predicting and simulating correlations between mechanical properties of age hardenable aluminium alloys as a function of alloy composition. Design/methodology/approach: Appropriate combinations of inputs and outputs were selected for neural network modelling. Multilayer feedforward networks were created and trained using datasets from public literature. Influences of alloying elements, alloy composition and processing parameters on mechanical properties of aluminium alloys were predicted and simulated using ANNs models.Two sample t-tests were used to analyze the prediction accuracy of the trained ANNs. Findings: Good performances of the neural network models were achieved. The models were able to predict mechanical properties within acceptable margins of error and were able to provide relevant simulated data for correlating alloy composition and processing parameters with mechanical properties. Therefore, ANNs models are convenient and powerful tools that can provide useful information which can be used to identify desired properties in new aluminium alloys for practical applications in new and/or improved aluminium products. Research limitations/implications: Few public data bases are available for modelling properties. Minor contradictions on the experimental values of properties and alloy compositions were also observed. Future work will include further development of simulated data into property charts. Practical implications: Correlations between mechanical properties and alloy compositions can help in identifying a suitable alloy for a new or improved aluminum product application. In addition, availability of simulated structure-process-property data or charts assists in reducing the time and costs of trial and error experimental approaches by providing near-optimal values that can be used as starting point in experimental work. Originality/value: Since the simulated data provides near-optimal values, manufacturers of new and/or improved aluminum alloys can use the simulated data as guidelines for narrowing down extensive experimental work. This in turn reduces the process design cycle times. Designers of new and/or improved aluminum products can also use the simulated data as a guideline for correlating property-application information, which is useful in preliminary design phase.

9

Reconstruction of the heat transfer coefficient on the grounds of experimental data

Słota D.

Journal of Achievements in Materials and Manufacturing Engineering

|

2009

|

Vol. 34, nr 1

63-70

EN

Purpose: Solidification of pure metal can be modelled by a two-phase Stefan problem, in which the distribution of temperature in the solid and liquid phases is described by the heat conduction equation with initial and boundary conditions. The inverse Stefan problem can be applied to solve design problems in casting process. Design/methodology/approach: In numerical calculations the alternating phase truncation method, the Tikhonov regularization and the genetic algorithm were used. The featured examples of calculations show a very good approximation of the experimental data. Findings: The verification of the method of reconstructing the cooling conditions during the solidification of pure metals. The solution of the problem consists of selecting the heat transfer coefficient on the boundary, so that the temperature in selected points on the boundary of the domain assumes given values. Research limitations/implications: The method requires that it must be possible to describe the sought boundary condition by means of a finite number of parameters. It is not necessary, that the sought boundary condition should be linearly dependent on those parameters. Practical implications: The presented method can be easy applied to solve design problems of different types, e.g. for the design of continuous casting installations (incl. the selection of the length of secondary cooling zones, the number of jets installed in individual zones, etc.). Originality/value: Verification, on the grounds of experimental data, the formerly devised method of determining the heat transfer coefficient during the solidification of pure metals.

10

Computer modelling system of the chemical composition and treatment parameters influence on mechanical properties of structural steels

Dobrzański L. A., Honysz R.

Journal of Achievements in Materials and Manufacturing Engineering

|

2009

|

Vol. 35, nr 2

138-145

EN

Purpose: This paper presents Neuro-Lab. It is an authorship programme, which use algorithms of artificial intelligence for structural steels mechanical properties estimation. Design/methodology/approach: On the basis of chemical composition, parameters of heat and mechanical treatment and elements of geometrical shape and size this programme has the ability to calculate the mechanical properties of examined steel and introduce them as raw numeric data or in graphic as influence charts. Possible is also to examine the dependence among the selected steel property and chosen input parameters, which describes this property. Findings: There is no necessity of carrying out any additional material tests. The results correlations between calculated and measured values are very good and achieve even the level of 98%. Practical implications: Presented programme can be an effective replace of the real experimental methods of properties determination in laboratory examinations. It can be applied as the enlargement of experimental work. Possible is also the investigation of models coming from new steel species, that wasn’t produced yet. Originality/value: The ability of the mechanical properties estimation of the ready, or foreseen to the use, material is unusually valuable for manufacturers and constructors. This signifies the fulfilment of customer’s quality requirements as well as measurable financial advantages for material manufacturers.

11

Metody regulacji kąta przegłębienia pojazdu podwodnego transportującego ładunek

Szymak P.

Polish Hyperbaric Research

|

2007

|

Nr 1(18)

17-31

PL

W przypadku wykorzystania pojazdu podwodnego do przenoszenia różnego rodzaju ładunków można zaobserwować efekt niepożądanego przegłębienia robota, co ma niekorzystny wpływ na proces sterowania jego ruchem. W referacie przedstawiono wyniki działania konwencjonalnych i rozmytych regulatorów kata przegłębienia pojazdu podwodnego, dostrajanych przy wykorzystaniu metod klasycznych oraz metod sztucznej inteligencji. Prace realizowano w Instytucie Podstaw Techniki Wydziału Mechaniczno-Elektrycznego Akademii Marynarki Wojennej w Gdyni.

EN

In the case of using an underwater vehicle to transfer different kind of loads, an effect of undesirable robot's trim might be observed, which has disadvantageous influence on a control process of its movement. In the paper, results of action of conventional and fuzzy underwater vehicle's trim controllers, tuned with the assistance of classical and artificial intelligence methods have been presented. Researches were carried out in The Electrotechnical and Electronic Department, The Naval University in Gdynia.

12

An improved region-growth algorithm for dense matching

Xing Y. J., Meng J. J., Sun J., Wang Z. Q.

Journal of Achievements in Materials and Manufacturing Engineering

|

2006

|

Vol. 18, nr 1-2

323--326

EN

Purpose: Improve the accuracy and speed of the region-growth algorithm between two 2D images. Design/methodology/approach: The algorithm includes two parts: the selection of seeds points and propagation. Some improvements are made in each one. For the first part, the best-first strategy is used to assure the accuracy of seeds. The epipolar line constraint and continuity constraint reduce the double phase matching course into single phase matching. For the second one, a dynamic and adaptive window is adopted instead of the large window. Findings: In the first section, the process of searching and the computational duties are decreased in large extent. And in the second one, the adaptive window makes the searching course more efficient in time and space. It is really difficult to get the most suitable window to search for the points as soon as possible. If it can be easily got, it will advance the efficiency of search. It is the future work. Practical implications: The method can be used in many different images, such as the structural images and the facial images. Originality/value: The original value is the region-growth algorithm, and in this paper I made some betterments to advance the efficiency and accuracy.