Person: DEREBAŞI, NAİM
Loading...
Email Address
Birth Date
8 results
Search Results
Now showing 1 - 8 of 8
Publication Experience of using a neural network for magnetic cores testing(Springer, 2021-03-08) Derebaşı, Naim; DEREBAŞI, NAİM; Bursa Uludağ Üniversitesi/Fen-Edebiyat Fakültesi/Fizik Bölümü,; 0000-0003-2546-0022; AAI-2254-2021Power loss in varied nanocrystalline toroidal cores was measured by using a modified and fully automated measuring system at induction frequency from 1 to 100 kHz and peak magnetic flux density from 0.1 to 1.0 T. Artificial neural network has been successfully used to analyse these collected data and predicted the power loss. In the developed model, the input parameters were outer and inner diameters, strip width, frequency and flux density, while the output parameter was the power loss. When the developed model was tested by untrained sample data, the average correlation of the model was found to be 99% and the overall prediction error was 0.23%. All models are developed with ANN and the results are in good agreement with the experimental results and they were within the acceptable limits.Publication Effect of geometrical factors on magnetic induction distribution of toroidal cores using numerical methods(Springer, 2015-03-01) Derebaşı, Naim; DEREBAŞI, NAİM; Uludağ Üniversitesi/Fen-Edebiyat Fakültesi/Fizik Bölümü; 0000-0003-2546-0022; AAI-2254-2021Toroidal cores wound from grain-oriented 3 % SiFe 0.27-mm thick (M4) electrical steel are widely used in many applications. Apart from the processing factors, the geometry of the toroid has an effect on its magnetic properties. In magnetic path length of a toroid from its inside to its outside circumference, the magnetic field varies at different layers. Artificial neural network have been successfully used for the prediction of magnetic performance in electromagnetic devices. Experimental data obtained from the previous measurements have been used as a training data to a feed forward multilayer perceptron neural network for the prediction of magnetic field and flux density distribution. The input parameters were outer and inner diameters and strip thickness while the output parameter was the magnetic field and flux density. When the network was tested by untrained sample data, the average correlation of the models was found to be 99 % and the overall prediction error was in the range of 2.31 and 0.02 in Ampere per meter and Tesla. An analytical equation as depending on theoretical data and prediction results has been determined by using MATLAB(a"double dagger) Curve Fitting Toolbox (TM) for magnetic field and flux density distribution in the toroid.Publication Influence of organic coating on the giant magneto impedance characteristics of fe-rich amorphous wire(Springer, 2015-03-01) Çaylak, Osman; Derebaşı, Naim; Çaylak, Osman; DEREBAŞI, NAİM; Uludağ Üniversitesi/Fen-Edebiyat Fakültesi/Fizik Bölümü; 0000-0001-8227-3456; 0000-0003-2546-0022; AAI-2254-2021; JAE-9301-2023Influence of organic coating on the giant magneto impedance effect was experimentally investigated in Zn complex-coated dielectric Fe-based amorphous wires and optimized the giant magneto impedance (GMI) effect using artificial neural networks and MATLAB. A three-node input layer, a one-node output layer and three hidden layers with 21 neurons and full connectivity between nodes were developed with the transfer functions hyperbolic tangent in hidden layers and sigmoid in output layer. The input parameters were frequency, static magnetic field and sample type, while the output parameter was the giant magneto impedance effect. When the network performance was tested using untrained sample data, the average correlation and prediction error of giant magneto impedance effect were found to be 99 and 0.4 %. An analytical equation as depending on experimental data has been determined by using MATLAB Curve Fitting Toolbox (TM) for giant magneto impedance. The square of the correlation and the root meansquared error were found to be 99 % and 0.89 respectively. The models including the different kinds of samples prepared have a good prediction capability and agreement with experimental results.Publication Performance of novel thermoelectric cooling module depending on geometrical factors(Springer, 2015-06-01) Derebaşı, Naim; Eltez, Muhammed; Güldiken, Fikret; Sever, Aziz; Kallis, Klaus; Kılıç, Halil; Özmutlu, Emin N.; DEREBAŞI, NAİM; Güldiken, Fikret; Özmutlu, Emin N.; Uludağ Üniversitesi/Fen-Edebiyat Fakültesi/Fizik Bölümü.; 0000-0003-2546-0022; AAI-2254-2021; CRO-8755-2022; FPR-2739-2022A geometrical shape factor was investigated for optimum thermoelectric performance of a thermoelectric module using finite element analysis. The cooling power, electrical energy consumption, and coefficient of performance were analyzed using simulation with different current values passing through the thermoelectric elements for varying temperature differences between the two sides. A dramatic increase in cooling power density was obtained, since it was inversely proportional to the length of the thermoelectric legs. An artificial neural network model for each thermoelectric property was also developed using input-output relations. The models including the shape factor showed good predictive capability and agreement with simulation results. The correlation of the models was found to be 99%, and the overall prediction error was in the range of 1.5% and 1.0%, which is within acceptable limits. A thermoelectric module was produced based on the numerical results and was shown to be a promising device for use in cooling systems.Publication Influence of geometrical factors on performance of thermoelectric material using numerical methods(Springer, 2015-06-01) Derebaşı, Naim; Eltez, Muhammed; Güldiken, Fikret; Sever, Aziz; Kallis, Klaus; Kılıç, Halil; DEREBAŞI, NAİM; Güldiken, Fikret; Uludağ Üniversitesi/Fen-Edebiyat Fakültesi/Fizik Bölümü; 0000-0003-2546-0022; AAI-2254-2021; CRO-8755-2022Prediction of the performance of thermoelectric cooling material (figure of merit, ZT) was carried out by simulated results obtained from the finite element method (FEM) as a training dataset with an artificial neural network. A total of 87 input vectors for the ZT obtained from the four thermoelectric cooling (TEC) modules modeled using the FEM analysis were available in the training set to a back-propagation artificial neural network. An average correlation and maximum prediction error were found to be 100% and 0.01%, respectively, for the ZT after training. The standard deviation of the values was 0.05%. A set of test data, different from the training dataset was used to investigate the network performance. The average correlation and maximum prediction error were found to be 99.92% and 0.07%, respectively, for the tested TEC module. A thermoelectric module produced based on the numerical results was shown to be a promising device for use in cooling systems.Publication Cooling performance of thermoelectric cooler modules: Experimental and numerical methods(Turkish Soc Thermal Sciences Technology, 2022-01-01) Kahraman, İlhan; Derebaşı, Naim; Kahraman, İlhan; DEREBAŞI, NAİM; Bursa Uludağ Üniversitesi/Fen-Edebiyat Fakültesi/Fizik Bölümü.; 0000-0003-2546-0022; 0000-0003-0662-1636; AAI-2254-2021; HHV-3379-2022A novel pulse-driving method in which the pulse frequency modulation is was developed by optimising the input power owing to the duty cycle of rectangular wave to enhance the cooling efficiency and thermal stability of the thermoelectric module. The aim of this driving method is to have better control of the thermoelectric cooler module temperature and to improve its coefficient of performance. In this method, the average current and the peak of pulse drive are in the 50% duty cycle with the same magnitude and the performance of Peltier module driving with average dc is compared with the pulse driving. The measurement results show that the coefficient of performance of the thermoelectric module with the pulse-frequency modulation driving method increased up to 102% as compared to the constant dc driving method. An artificial neural network has been successfully used to analyse these experimentally collected data and predict the performance of the module. When the developed artificial neural network model was tested using untrained data, the average correlation of the model was 99% and the overall prediction error was 1.38%. An accurate and simple analytical equation based on the predicted and experimental results was determined using the MATLAB (R) Curve Fitting Toolbox. The average correlation of the analytical model was 0.99 and the root-mean-square error was 0.074.Publication Prediction of optical parameters of sn doped cdo films using neural network(Natl Inst Optoelectronics, 2008-02-01) Köse, S.; Atay, F.; Bilgin, V.; Akyuz, I.; Ertürk, Kadir; Haciismailoglu, M. C.; HACIİSMAİLOĞLU, MUHAMMED CÜNEYT; Küçük, İ.; Derebaşı, Naim; DEREBAŞI, NAİM; Bursa Uludağ Üniversitesi/Fen Edebiyat Fakültesi/Fizik Bölümü.; 0000-0001-5650-9146; 0000-0002-0880-5028; 0000-0002-0781-3376; 0000-0001-8483-7366; 0000-0003-2546-0022; AAG-5509-2019; K-7950-2012; ABG-7537-2020; A-1120-2010; AAV-3055-2021; ABA-5148-2020; AAI-2254-2021In recent years, there was great interest and demand for the production and investigation of low cost and novel transparent conducting oxide films. CdO is a promising material among these films for future applications with its unique properties. A learning and generalization ability, real-time operation, and ease of implementation have made an artificial neural network popular in recent years. In this work we have produced CdO:Sn films by the ulrasonic spray pyrolysis technique which is economical and simple to process. Optical parameters of Sn doped CdO films with developed, have been estimated by the artificial neural network using experimental results as a training data. The correlation obtain from the artificial neural network was found to be 99% with the experimental results.Publication Modelling of power loss in electrical steels(Polish Acad Sciences Inst Physics, 2008-01-01) Küçük, İ.; Ertürk, K.; Haciismailoğlu, M. C.; HACIİSMAİLOĞLU, MUHAMMED CÜNEYT; Derebaşı, Naim; DEREBAŞI, NAİM; Bursa Uludağ Üniversitesi/Fen Edebiyat Fakültesi/Fizik Bölümü.; 0000-0002-0781-3376; 0000-0003-2546-0022; ABA-5148-2020; AAI-2254-2021; K-7950-2012This paper presents a new artificial neural network approach based on loss separation model to compute power loss on different types of electrical steels. The network was trained by a Levenberg-Marquardt algorithm. The results obtained by using the proposed model were compared with a commonly used conventional model. The comparison has shown that the neural network model is in good agreement with experimental data with respect to the conventional model.