2/2020
• 2/2020 table of contents
NEURAL NETWORKS FOR LEARNING INVERSE KINEMATICS OF ROBOTS: REVIEW AND APPLICATION 5
Hazim Nasir Ghafil, Károly Jármai
Deep learning for Inverse Kinematic problem of robot manipulators was investigated by many works considering the planar case. In this paper, the performance of the three learning algorithms; Levenberg-Marquardt algorithm, Bayesian Regularization algorithm, and Scaled Conjugate Gradient algorithm is studied will learning single hidden layer network how to solve the Inverse Kinematic problem of a three degree of freedom robot manipulator.
BEAM WITH FIXED SUPPORT AT ONE END OPTIMIZATION IN CASE OF ECCENTRIC LOAD 11
Antal Erdős, Károly Jármai
In this article the optimization of a beam with opened cross section. is shown considering the stress and deflection calculations. This type of cross-section is very sensitive to the eccentricity of the load, because of the effect of the warping torsion. This type of model could be used in case of machine tools, for example press or drill machines. Because the accuracy of the manufacturing procedure and the workpiece the deflection of the beam must be minimal. On the other hand, the lifetime of the machine should be maximal.
OPTIMUM DESIGN OF WELDED ASYMMETRIC I-BEAMS FOR MIMIMUM WELDING SHRINKAGE 19
Károly Jármai, Máté Petrik
A calculation system has been developed to determine the optimum dimensions of asymmetric I-beams for minimum shrinkage. The objective function is the minimum mass, the unknowns are the I-beam dimensions, the constraint are the stress, local buckling and deflection. We have
considered different steel grades (235, 355, 460 MPa yield stress) and different aluminium (90, 155, 240 MPa yield stress). The material, the span length and the loading have been changed. It is shown, that using optimum design, one can reduce the welding shrinkage with prebending and can safe material cost as well.
OPTIMIZATION OF OVERHEAD TRAVELLING CRANE WITH FPA ALGORITHM ON GPU 27
Szilárd Nagy, Károly Jármai
Nature-inspired evolutionary optimization algorithms are powerful tools for solving non-linear problems. Sometimes they require huge computation capability, and they may be slow. In this paper, we propose a possible parallelization method for computation of base FPA algorithm and one group of fitness function. Proposed method simulated with three test
function and an optimization of main girder of overhead travelling crane.
PUBLICATION AND MODERN SEARCH OF THE LITERATURE: INTRODUCTION OF USING SCOPUS 33
Viktória Kállai, Károly Jármai
The Scopus is a database, which has the largest collection of abstracts and citations. Furthermore, with different intelligent tools it makes possible the observation of researches. Using Scopus one can not only search and find relevant information about the searched topic, but it is possible to find research groups, and to make statistical analysis
of articles, authors or journals. It offers a huge amount and comprehensive content at one source. Information related to all disciplines can be searched easily and efficiently. It provides access to thousands of articles, millions of author profiles and billions of
citations cited.
ANALYTICAL AND NUMERICAL INVESTIGATION OF A SANDWICH BEAM WITH ALUMINIUM FOAM 40
Alaa Al-Fatlawi, Károly Jármai, György Kovács
The purpose of the research is to develop a new lightweight sandwich structure that can be used as element of air containers. The structure is made of aluminium foam with fiber reinforced composite cover plates. Nine different combinations of laminated glass and/or carbon fiber reinforced plastic sheets have been tested. A finite element analysis of the sandwich structures was presented. The single-objective optimization of the new sandwich structure was achieved with minimal weight. Five design criteria have been considered: stiffness of the structure, shear failure of the topsheet, shear of the core, wrinkling of the topsheet, size limitations of the design variables. The elaborate composite structure results in significant weight savings due to the low density.
NEURAL NETWORKS FOR LEARNING INVERSE KINEMATICS OF ROBOTS: REVIEW AND APPLICATION 5
Hazim Nasir Ghafil, Károly Jármai
Deep learning for Inverse Kinematic problem of robot manipulators was investigated by many works considering the planar case. In this paper, the performance of the three learning algorithms; Levenberg-Marquardt algorithm, Bayesian Regularization algorithm, and Scaled Conjugate Gradient algorithm is studied will learning single hidden layer network how to solve the Inverse Kinematic problem of a three degree of freedom robot manipulator.
BEAM WITH FIXED SUPPORT AT ONE END OPTIMIZATION IN CASE OF ECCENTRIC LOAD 11
Antal Erdős, Károly Jármai
In this article the optimization of a beam with opened cross section. is shown considering the stress and deflection calculations. This type of cross-section is very sensitive to the eccentricity of the load, because of the effect of the warping torsion. This type of model could be used in case of machine tools, for example press or drill machines. Because the accuracy of the manufacturing procedure and the workpiece the deflection of the beam must be minimal. On the other hand, the lifetime of the machine should be maximal.
OPTIMUM DESIGN OF WELDED ASYMMETRIC I-BEAMS FOR MIMIMUM WELDING SHRINKAGE 19
Károly Jármai, Máté Petrik
A calculation system has been developed to determine the optimum dimensions of asymmetric I-beams for minimum shrinkage. The objective function is the minimum mass, the unknowns are the I-beam dimensions, the constraint are the stress, local buckling and deflection. We have
considered different steel grades (235, 355, 460 MPa yield stress) and different aluminium (90, 155, 240 MPa yield stress). The material, the span length and the loading have been changed. It is shown, that using optimum design, one can reduce the welding shrinkage with prebending and can safe material cost as well.
OPTIMIZATION OF OVERHEAD TRAVELLING CRANE WITH FPA ALGORITHM ON GPU 27
Szilárd Nagy, Károly Jármai
Nature-inspired evolutionary optimization algorithms are powerful tools for solving non-linear problems. Sometimes they require huge computation capability, and they may be slow. In this paper, we propose a possible parallelization method for computation of base FPA algorithm and one group of fitness function. Proposed method simulated with three test
function and an optimization of main girder of overhead travelling crane.
PUBLICATION AND MODERN SEARCH OF THE LITERATURE: INTRODUCTION OF USING SCOPUS 33
Viktória Kállai, Károly Jármai
The Scopus is a database, which has the largest collection of abstracts and citations. Furthermore, with different intelligent tools it makes possible the observation of researches. Using Scopus one can not only search and find relevant information about the searched topic, but it is possible to find research groups, and to make statistical analysis
of articles, authors or journals. It offers a huge amount and comprehensive content at one source. Information related to all disciplines can be searched easily and efficiently. It provides access to thousands of articles, millions of author profiles and billions of
citations cited.
ANALYTICAL AND NUMERICAL INVESTIGATION OF A SANDWICH BEAM WITH ALUMINIUM FOAM 40
Alaa Al-Fatlawi, Károly Jármai, György Kovács
The purpose of the research is to develop a new lightweight sandwich structure that can be used as element of air containers. The structure is made of aluminium foam with fiber reinforced composite cover plates. Nine different combinations of laminated glass and/or carbon fiber reinforced plastic sheets have been tested. A finite element analysis of the sandwich structures was presented. The single-objective optimization of the new sandwich structure was achieved with minimal weight. Five design criteria have been considered: stiffness of the structure, shear failure of the topsheet, shear of the core, wrinkling of the topsheet, size limitations of the design variables. The elaborate composite structure results in significant weight savings due to the low density.
view the complete magazine, download pdf
