Search

Copyright of Duzce University Journal of Science & Technology is the property of Duzce University Journal of Science & Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

• The performance of counter-flow Ranque-Hilsch Vortex Tube (RHVT) was modelled with respect to pressure, working fluid and nozzle specifications. • Linear Regression (LR), Support Vector Machines (SVM), Gaussian Process Regression (GPR), Regression Trees (RT) and Ensembles of Trees (ET) prediction methods were used. • Optimizing the performance of counter-flow RHVT, it is aimed to fill the gap in the literature by using LR, SVM, GPR, RT and ET methods among the machine learning methods. This study planned to model a counter-flow Ranque-Hilsch Vortex Tube (RHVT) using compressed air and oxygen gas by machine learning to separate the thermal temperature. From within machine learning models, Linear Regression (LR), Support Vector Machines (SVM), Gaussian Process Regression (GPR), Regression Trees (RT), and Ensemble of Trees (ET) were preferred. By leaving the outlet control valve on the hot fluid side fully open, data were received for each material and nozzle at RHVT with inlet pressure starting from 150 kPa and up to 700 kPa at 50 kPa intervals. In the counter flow RHVT, the lack in the literature has been tried to be eliminated by modeling the RHVT by finding the difference (ΔT) between the temperature of the cold flow exiting (T c) and the temperature of the leaving hot flow (T h). When analyzing each of the machine learning models in the study, 80% of all data was used as training data, 20% of all data was used for the test, 70% of all data was used as training data, and 30% of all data was used for the test. As a result of the analysis, when both air and oxygen fluids were used, the GPR method gave the best result with 0.99 among the machine learning models in two different test intervals of 70%–30% and 80%–20%. The success of other machine learning models differed according to the fluid and model used. [ABSTRACT FROM AUTHOR]

Abstract: Freeze drying is the best drying technology regarding quality of the end product but it is an expensive method and the high costs of process limit its application to industrial scale. At the same time, the freeze-drying process is based on different parameters, such as drying time, pressure, sample thicknesses, chamber temperature, sample temperatures and relative humidity. So, the determination of drying behaviors, such as moisture content (MC), moisture ratio (MR) and drying rate (DR), of the freeze-drying process are too complex. In this paper, to help the freeze dryer designer and simplify this complex process, the use of artificial neural networks has been proposed. An artificial neural networks (ANN) model has been developed for determination the prediction of drying behaviors, such as MC, MR and DR, of apples in the freeze-drying process. The back-propagation learning algorithm with variant which is Levenberg–Marquardt (LM) and Fermi transfer function have been used in the network. In addition, the statistical validity of the developed model has been determined by using the coefficient of determination (R 2), the root means square error (RMSE) and the mean absolute percentage error (MAPE). R 2, RMSE and MAPE have been determined for MC, MR and DR, as 0.999, 0.0078895, 0.2668459, and 0.999, 0.0001099, 0.2968427 and 0.999, 0.0000008, 0.2703797, respectively. [ABSTRACT FROM AUTHOR]