Research Outputs
Permanent URI for this communityhttps://hdl.handle.net/20.500.14288/2
Browse
8 results
Search Results
Publication Open Access A systematic and efficient input selection method for artificial neural networks using mixed-integer nonlinear programming(Konya Teknik Üniversitesi, 2022) Şıldır, Hasan; Department of Chemical and Biological Engineering; Aydın, Erdal; Faculty Member; Department of Chemical and Biological Engineering; College of Engineering; 311745Selection of input variables of the empirical models has vital effect on the prediction performance, reduced overfitting and reduced computational load. Various trials and error and sequential methods in the literature to deal with input selection for artificial neural networks (ANNs). However, these methods are not considered as automatic and systematic. This study proposes a novel and efficient mixed integer nonlinear programming-based approach to handle optimal input selection and the ANN training simultaneously for classification problems. Such selection uses binary (0-1) variables to represent the presence of the input variables and trains traditional continuous network weights simultaneously. Two classification case studies are given to demonstrate the advantages by using widely used data sets and statistical measures. The first data set is related to the characterization of the type of a tumor related to breast cancer, the second data set is about predicting the type of a biotechnological product using different features, the last one is related to heart failure prediction. Results show that better test performance can be achieved with optimally selected inputs, resulting in reduced overfitting. The proposed approach delivers a significant advantage during the design and training of the ANNs and is also applicable to other empirical models. / Ampirik modellerin girdi değişkenlerinin seçimi, tahmin performansı, azaltılmış fazla uydurma ve hesaplama yükünün azaltılması üzerinde önemli etkiye sahiptir. Literatürde yapay sinir ağları (YSA) için girdi seçimi ile ilgili çeşitli deneme yanılma yöntemleri mevcuttur ancak bu metodlar sistematik ve otomatik olarak kabul edilmemektedir. Bu çalışma, sınıflandırma problemleri için optimal girdi seçimi ve YSA eğitimini aynı anda ele almak için yeni ve verimli bir karma tamsayılı doğrusal olmayan programlama tabanlı bir yaklaşım önermektedir. Bu seçim, girdi değişkenlerinin varlığını temsil etmek için ikili (0-1) değişkenleri kullanır ve geleneksel sürekli ağ ağırlıklarını veya parametrelerini aynı anda eğitir. Yaygın olarak kullanılan veri setleri ve istatistiksel ölçümler kullanarak avantajları göstermek amacıyla üç sınıflandırma vaka çalışması sunulmuştur. Birinci veri seti meme kanseri ile ilgili tümörün tipin-in karakterizasyonu ile ilgili olup, ikinci veri seti ise farklı özellikler kullanılarak bir biyoteknolojik ürünün tipinin tahmin edilmesi ile ilgilidir, son veri seti ise kalp sağlığı ile ilgilidir. Sonuçlar, optimal olarak seçilen girdiler ile düşük fazla uydurma sayesinde daha iyi test performansının elde edilebileceğini göstermektedir. Önerilen yaklaşım, YSA'ların tasarımı ve eğitimi sırasında önemli bir avantaj sağlar ve diğer ampirik modellere de uygulanabilir.Publication Open Access CO2 absorption into primary and secondary amine aqueous solutions with and without copper ions in a bubble column(TÜBİTAK, 2022) Güler, Cansu; Uzunlar, Erdal; Department of Chemical and Biological Engineering; Erkey, Can; Yousefzadeh, Hamed; Faculty Member; Researcher; Department of Chemical and Biological Engineering; College of Engineering; 29633; N/AChemical absorption of CO2 into aqueous amine solutions using a nonstirred bubble column was experimentally investigated. The performance of CO2 absorption of four different primary and secondary amines including monoethanolamine (MEA), piperazine (PZ), 2-piperidineethanol (2PE), and homopiperazine (HPZ) were compared. The effects of initial concentration of amine, the inlet mole fraction of CO2, and solution temperature on the rate of CO2 absorption and CO2 loading (mol CO2/mol amine) were studied in the range of 0.02–1 M, 0.10–0.15, and 25–40 °C, respectively. The effect of the presence of copper ions in the amine solution on CO2 loading was also studied. By comparison of the breakthrough curves of the amines at different operational conditions, it was revealed that the shortest and longest time for the appearance of the breakthrough point was observed for MEA and HPZ solutions, respectively. CO2 loading of MEA, 2PE, PZ, and HPZ aqueous solutions at 25 °C, 0.2 M of initial concentration of amine, and 0.15 of inlet mole fraction of CO2 were 1.06, 1.14, 1.13, and 1.18 mol CO2/mol amine, respectively. By decreasing the inlet mole fraction of CO2 from 0.15 to 0.10, CO2 loading slightly decreased. As the initial concentration of amine and temperature decreased, CO2 loading increased. Also, the presence of copper ions in the absorbent solution resulted in a decrease in the CO2 loading of MEA and HPZ aqueous solutions. In case of PZ and 2PE amines, adding copper ions led to precipitation even at low copper ion concentrations.
Publication Metadata only Competitive hydrogen bonding in aspirin-aspirin and aspirin-leucine interactions(Scientific Technical Research Council Turkey-Tubitak, 2012) Department of Chemistry; Department of Chemical and Biological Engineering; Department of Chemistry; Yurtsever, Zeynep; Erman, Burak; Yurtsever, İsmail Ersin; Undergraduate Student; Faculty Member; Faculty Member; Department of Chemical and Biological Engineering; Department of Chemistry; College of Sciences; College of Engineering; College of Sciences; N/A; 179997; 7129Aspirin-aspirin and aspirin-leucine interactions are studied by the density functional theory (DFT) and high level ab initio calculations with second order Moller-Plesset perturbation theory (MP2). The rotational isomers of aspirin are identified by their relative stability both in gaseous phase and in water using the polarizable continuum method (PCM). Local minima of aspirin monomers in water are found to be all highly populated compared to the gas phase behavior. Homodimers of aspirin form hydrogen bonds with bond energies of 10 kcal/mol. Weak hydrogen bonds utilizing phenyl and methyl groups are also found. The interaction between aspirin and leucine is stronger with relatively short bond lengths compared to homodimeric aspirin interactions. The potential energy surface has several minima with comparable stability. This study shows the significance of diverse bonding schemes, which are important for understanding complete interaction mechanisms of aspirin.Publication Metadata only Enhanced ionic conductivity and mechanical strength in nanocomposite electrolytes with nonlinear polymer architectures(TÜBİTAK, 2023) N/A; Department of Chemical and Biological Engineering; N/A; Bakar, Recep; Şenses, Erkan; Darvishi, Saeid; PhD Student; Faculty Member; PhD Student; Department of Chemical and Biological Engineering; Koç University Boron and Advanced Materials Application and Research Center (KUBAM) / Koç Üniversitesi Bor ve İleri Malzemeler Uygulama ve Araştırma Merkezi (KUBAM); Koç University Surface Science and Technology Center (KUYTAM) / Koç Üniversitesi Yüzey Teknolojileri Araştırmaları Merkezi (KUYTAM); Graduate School of Sciences and Engineering; College of Engineering; Graduate School of Sciences and Engineering; N/A; 280298; N/ASolvent-free polymer-based electrolytes (SPEs) have gained significant attention to realize safer and flexible lithium-ion batteries. Among all polymers used for preparing SPEs electrolytes, poly(ethylene oxide), a biocompatible and biodegradable polymer, has been the most prevalent one mainly because of its high ionic conductivity in the molten state, the capability for the dissolution of a wide range of different lithium salts as well as its potential for the environmental health and safety. However, linear PEO is highly semicrystalline at room temperature and thus exhibits weak mechanical performance. Addition of nanoparticles enhances the mechanical strength and effectively decreases the crystallization of linear PEO, yet enhancement in mechanical performance often results in decreased ionic conductivity when compared to the neat linear PEO-based electrolytes; new strategies for decoupling ionic conductivity from mechanical reinforcement are urgently needed. Herein, we used lithium bis(trifluoromethane-sulfonyl)-imide (LiTFSI) salts dissolved in various nonlinear PEO architectures, including stars (4-arms and 8-arms) and hyperbranched matrices, and SiO2 nanoparticles (approximately equal to 50 nm diameter) as fillers. Compared to the linear PEO chains, the room temperature crystallinity was eliminated in the branched PEO architectures. The electrolytes with good dispersion of the nanoparticles in the nonlinear PEOs significantly enhanced ionic conductivity, specifically by approximately equal to 40% for 8-arm star, approximately equal to 28% for 4-arms star, and approximately equal to %16 for hyperbranched matrices, with respect to the composite electrolyte with the linear matrix. Additionally, the rheological results of the SPEs with branched architectures show more than three orders of magnitude enhancement in the low-frequency moduli compared to the neat linear PEO/Li systems. The obtained results demonstrate that the solvent-free composite electrolytes made of branched PEO architectures can be quite promising especially for irregularly shaped and environmentally benign battery applications suitable for medical implants, wearable devices, and stretchable electronics, which require biodegradability and biocompatibility. © TÜBİTAK.Publication Metadata only In silico design of novel and highly selective lysine-specific histone demethylase inhibitors(Scientific Technical Research Council Turkey-Tubitak, 2011) Akdogan, Ebru Demet; Yelekci, Kemal; Department of Chemical and Biological Engineering; Erman, Burak; Faculty Member; Department of Chemical and Biological Engineering; College of Engineering; 179997Histone lysine-specific demethylase (LSD1) is involved in a wide range of epigenetic processes and plays important roles in gene silencing, DNA transcription, DNA replication, DNA repair, and heterochromatin formation. Its active site shows a resemblance to those of 2 homologous enzymes, monamine oxidase A and B (MAO-A and MAO-B.) In the present work, starting from suitable scaffolds and generating thousands of structures from them, 10 potential inhibitors were obtained with structural and physicochemical properties selectively suitable for inhibiting LSD1. iLib Diverse software was used to generate the diverse structures and 3 docking tools, CDOCKER, GOLD, and AutoDock, were used to find the most probable potential inhibitor based on its binding affinity. The dispositions of the candidate molecules within the organism were checked by ADMET_PSA_2D (polar surface area) versus ADMET_AlogP98 (the logarithm of the partition coefficient between n-octanol and water), and their suitability is discussed. The LSD1 inhibition activities of the candidates were compared with the properties of trans-2-phenylcyclopropylamine (tranylcypromine) and 2-(4-methoxy-phenyl) cyclopropylamine, which are the 2 known inhibitors of LSD1.Publication Metadata only Numerical investigation of design parameters effects on performance of cooling system designed for a lithium-ion cell(Yildiz Technical University, 2020) N/A; Department of Chemical and Biological Engineering; Alipour, Mohammad; Kızılel, Rıza; PhD Student; Researcher; Department of Chemical and Biological Engineering; Graduate School of Sciences and Engineering; College of Engineering; N/A; 114475A 3D numerical approach using the Finite Element Method (FEM) is applied to model the thermal behavior of multilayer 20Ah LiFePO4/Graphite cell and to design a cooling system. A three-dimensional multilayer cell model with heterogeneous thermal properties for the various cell layers is developed to study the effects of design parameters on cooling performance of mini-channel aluminum plates. As design parameters, effects of channel width, a number of channel passes, inlet mass flow rate, and heat transfer medium were considered. Using the optimized parameters, the cooling performance of water-cooling and air-cooling systems were compared. The results showed that the designed cooling system provided good cooling performance in controlling the temperature rise and uniformity. Inlet mass flow rate was the main influential parameter in controlling the cooling performance. The optimum number of channel passes was found to be seven passes. Channel width mainly controlled the pressure drop and had minor effects on temperature. At higher discharge current rates, the water-cooling system showed better cooling performance in dropping the maximum temperature and making uniform surface and inner temperature profile. Moreover, pressure drop, and power consumption rates become significantly lower for water cooling system.Publication Open Access Reaction network reduction with mixed-integer nonlinear programming(Konya Teknik Üniversitesi, 2021) Şıldır, Hasan; Ertürk, Emrullah; Department of Chemical and Biological Engineering; Aydın, Erdal; Faculty Member; Department of Chemical and Biological Engineering; College of Engineering; 311745In this study, a Mixed-Integer Nonlinear programming (MINLP) problem is formulated for reaction network model reduction. The MINLP problem introduces binary variables for the existence of rate constants in addition to traditional continuous variables to minimize the prediction error. Such binary variables are implemented through linking constraints. Both the impact of initial conditions and operating conditions are investigated on the model reduction. Commercial and free solver comparisons are also provided in terms of computational time and results. The methodology is implemented on an experimentally-derived reaction pathway from the literature. A significant network reduction is achieved under different operating temperatures and initial conditions. The reduced model provides a satisfactory prediction accuracy with significantly low number of reactions and parameters. / Bu çalışmada, reaksiyon ağı küçültmesi için tam sayılı ve kesikli bir optimizasyon (MINLP) problemi formüle edilmiştir. Bu problem, tahmin hatasını enküçüklemek için geleneksel sürekli değişkenlere ek olarak reaksiyon hız sabitlerinin mevcudiyeti için iki değerli değişkenler tanımlamaktadır. Bu iki değerli değişkenler bağlantı kısıtı ile uygulanmaktadır. Başlangıç koşulları ve çalışma koşullarının model küçültmeye olan etkisi araştırılmıştır. Bu bağlamda, ticari ve ücretsiz çözücü programların hesaplama süreleri ve sonuçları karşılaştırmalı olarak sunulmuştur. Önerilen yöntem literatürde bulunan deneysel olarak türetilmiş reaksiyon ağına uygulanmıştır. Farklı sıcaklık ve başlangıç konsantrasyonlarında kayda değer ağ küçültülmesi sağlanmıştır. Küçültülmüş model önemli ölçüde az reaksiyon ve parametre sayısı ile tatmin edici kestirim doğruluğu sunmaktadır.Publication Metadata only Single and multisite detailed kinetic models for the adsorption and desorption of NO2 over Cu based NH3-SCR catalyst(Mersin Üniversitesi, 2022) Department of Chemical and Biological Engineering; Bozbağ, Selmi Erim; Researcher; Department of Chemical and Biological Engineering; College of Engineering; N/AKinetic modeling of NH3 Selective Catalytic Reduction (NH3-SCR) of NOx in Cu-chabazite washcoated monolithic reactors has recently become an important task for design, control and calibration of heavy-duty engine aftertreatment systems. Development of detailed and accurate kinetic models rely on the correct simulation of the NO2 and NH3 storage at different conditions. Here, different kinetic schemes for NO2 adsorption and desorption were developed and compared to experimental data. For this purpose, firstly, realistic values of the active Cu sites in the Cu-zeolite were obtained using the temperature programmed desorption (TPD) of NH3 and NO2 which showed fractional coverages of 0.04 and 0.17 for the so-called ZCuOH and Z2Cu species which reside in the 8 and 6 membered rings (MR) of the zeolitic framework, respectively. Active site concentrations were used in the kinetic models which included simultaneous formation of nitrate/nitrite species or the formation of HNO3 intermediate which in turn resulted in the formation of nitrates or nitrites over the ZCuOH. Models also included or excluded the NO2 storage over the so called secondary Z2Cu sites. It was shown that models taking into account HNO3 intermediate formation along with two NO2 storage sites were better fits to the experimental data.