Yazar "Goz, Eda" seçeneğine göre listele
Listeleniyor 1 - 7 / 7
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe BROMATE REMOVAL PREDICTION IN DRINKING WATER BY USING THE LEAST SQUARES SUPPORT VECTOR MACHINE (LS-SVM)(Yildiz Technical Univ, 2020) Karadurnius, Erdal; Goz, Eda; Taskin, Nur; Yuceer, MehmetThe main objective of this study was to develop Least Squares Support Vector Machine (LS-SVM) algorithm for prediction of bromate removal in drinking water. Adsorption method known as environmental-friendly and economical was used in the experimental part of this study to remove this harmful compound from drinking water. Technically (pure), HCl-, NaOH- and NH3-modified activated carbons were prepared as adsorbent. Experimental studies were carried out with synthetic samples in three different concentrations. To forecast bromate removal percentage particle size and amount of the activated carbon, height and diameter of the column, volumetric flowrate, and initial concentration were selected as the input variables Radial basis kernel function was selected as activation function in algorithm. Algorithm parameters that gamma and sigma(2) values set as 415 and 3.956 respectively. To evaluate model performance some performance indices were calculated. Correlation coefficient (R), mean absolute percentage error (MAPE%) and root mean square error (RMSE) value for the training and testing phase R:0.996, MAPE%: 2.59 RMSE: 2.14 and R:0.994, MAPE%: 3.21 RMSE: 2.51 respectively. These results obtained from this study were compared with the ANN model previously developed with the same input data. As a result, LS-SVM has better performance than ANN.Öğe COMPARISON OF CONTROL STRATEGIES FOR DISSOLVED OXYGEN CONCENTRATION IN ACTIVATED SLUDGE PROCESS(Parlar Scientific Publications (P S P), 2016) Akyurek, Evrim; Karadurmus, Erdal; Yuceer, Mehmet; Goz, Eda; Atasoy, Ilknur; Berber, RidvanDifferent control algorithms were compared and tested for activated sludge wastewater treatment process. Proportional-integral-derivative control (PID), Model Predictive Control (MPC) with linear model, MPC with non-linear model, Nonlinear Autoregressive-Moving Average (NARMA-L2) control, Neural Network Model Predictive Control (NN-MPC) and optimal control with Sequential Quadratic Programming (SQP) algorithm were evaluated via simulation of activated sludge model. Controlled and manipulated variables were selected as dissolved oxygen level and aeration rate, respectively. Rise time, overshoot, Integral Absolute Error (IAE) and Integral Square Error (ISE) were calculated for each controller. It was concluded that NARMA-L2 controller and optimal control with SQP would outperform the other control strategies.Öğe FRACTIONAL-ORDER CONTROL STRATEGIES FOR THE ACTIVATED SLUDGE PROCESS(Parlar Scientific Publications (P S P), 2018) Goz, Eda; Yuceer, MehmetActivated sludge process has a complex and nonlinear characteristics, therefore, various conventional control algorithms are incapable of controlling activated sludge process. On the other hand, optimizations of waste water treatment plants have been inevitable due to the strict regulations. This paper deals with the application of fractional order proportional integral and derivative controller ((PID mu)-D-lambda), fractional order proportional integral (PI lambda) and classical PID and PI controller for activated sludge waste water treatment plants. For this purpose, a simpler model with single aeration bioreactor that includes biological process with one type of substrate and microorganism was used. Since the level of dissolved oxygen in the aeration tank is important for the effluent quality standards and minimizing the operating cost, it was chosen as the controlled variable. Moreover, manipulated variable has been defined as aeration rate. Performances of PID, PI, fractional order PID and fractional order PI controller were evaluated with integral square error (ISE). The simulation results indicated that fractional order PID controller exhibited a better performance than fractional order PI, conventional PID and PI controllers. Controller parameters were calculated via various optimization strategies such as particle swarm optimization (PSO), genetic algorithm (GA) and sequential quadratic programming (SQP). The values of fractional order PID and PI controller parameters were almost the same even if different optimization methods were tried for determining the controller parameters. Fractional order PID and fractional order PI controller parameters were obtained as K-p = 29.8916, K-I = 29.913, K-D = 29.8909, lambda = 1.0934, mu = 0.01612 and K-p = 29.9999, K-I = 29.9999, lambda = 0.995, respectively. Similarly, conventional PID parameters were not affected by different optimization methods. Both GA and PSO show the same controller parameters (K-p = 29.9997, tau(I) = 29.9998, tau(D) = 29.9993). Furthermore, conventional PID parameters that are manipulated by Ziegler Nichols method were K-p = 46, tau(I) = 115, tau(D) = 4.6.Öğe Kinetic and artificial neural network modeling techniques to predict the drying kinetics of Mentha spicata L.(Wiley, 2019) Karakaplan, Nihan; Goz, Eda; Tosun, Emir; Yuceer, MehmetThis study presented both the empirical and artificial neural network (ANN) approaches to estimate the moisture content of Mentha spicata. Two different types of drying methods (in shade and in oven (35 and 50 degrees C)) were used to investigate the drying kinetics of the Mentha spicata samples. The effects of drying methods on effective diffusion coefficient, moisture ratio (MR), drying rate, and activation energy were investigated. Moreover, six different thin layer drying models (Page, Diffusion approach, Newton, Modified Henderson, Henderson and Pabis and Pabis and Midilli) and an ANN with feed forward structure were used to define the drying kinetics of these samples. In order to estimate the kinetic model parameters, sequential quadratic programming (SQP) was used. Model performances were evaluated based on the coefficient of determination (R-2), root mean square error (RMSE) and mean absolute percentage error (MAPE%) values. In the kinetic part of the modeling study, the Midilli model provided better results than the others. However, the ANN had the best results when a total assessment was made. The effective diffusion coefficient values were found in the range between 1.31 x 10(-12) and 4.43 x 10(-12) m(2)/s. The activation energy was obtained as 44.31 kJ/kmol. The R-2, MAPE%, and RMSE values for the ANN test data were 1.00, 0.2257, and 5.9447 x 10(-4), respectively. In the future, different modeling approaches will be applied to describe this drying process. Practical applications Drying is a process where heat transfer and mass transfer take place together. Modeling is an innovative approach used in evaluation of experimental data and has increasing popularity in recent years. ANNs are a powerful data-driven method, and they have a very broad area of usage from medicine to engineering issues. Empirical models are another approach for describing experimental data. In this study, these two modeling approaches were used to obtain the MR. Humidity is a condition that needs to be checked in food safety and protection. Therefore, it is very important to ensure control with robust modeling techniques. In this study, the developed ANN model had a high R-2 value (R-2 = 1.00). This indicated that it may be used successfully in real applications.Öğe Prediction of Bromate Removal in Drinking Water Using Artificial Neural Networks(Taylor & Francis Inc, 2019) Karadurmus, Erdal; Taskin, Nur; Goz, Eda; Yuceer, MehmetIn treatment of natural water resources, bromide transforms into carcinogenic bromate, especially during the ozonation process. Adsorption was used in the experimental part of this study to remove this harmful compound from drinking water. For this purpose, technically, HCl-, NaOH-, and NH3-modified activated carbons were used. Scanning Electron Microscopy (SEM) and Brunauer-Emmett-Teller (BET) analyses were carried out within the characterization study. Moreover, the effects of diameters and heights of adsorption columns, flowrate, and particle size of adsorbent were investigated on the removal amounts of bromate. Optimum conditions were obtained from the experiments, and regional/real samples were collected and analyzed. After the experiments, an artificial neural network (ANN) was used to predict bromate removal percentage by using the observed data. Within this context, a feed-forward back-propagation ANN was chosen in this study. Additionally, the transfer function was selected as tangent sigmoid and 3 neurons were used in the hidden layer. Particle size and amount of the activated carbon, height and diameter of the column, volumetric flowrate, and initial concentration were selected as the input variables. Bromate removal percentage was selected as the output. It was found that the model an R value of 0.988, RMSE value of 3.47 and mean absolute percentage error (MAPE) of 5.19% in the test phase.Öğe Prediction of characteristic properties of crude oil blending with ANN(Taylor & Francis Inc, 2018) Karadurmus, Erdal; Akyazi, Habib; Goz, Eda; Yuceer, MehmetMineral oil is one of the most important materials on earth and it is used widely for its several features. Mineral oils derived from petroleum products are commonly used to decrease the friction effects in machine parts and, thus, they both prevent wear/overheating and facilitate power transmission. In this study, various binary mixtures of various base oils (SN-80, SN-100, SN-150, SN-50, SN-500) were prepared at different volumetric ratios. Kinematic viscosity (at 40 degrees C and 100 degrees C), viscosity index, flash point, pour point, and density (at 20 degrees C) measurements were performed for characterization of the prepared mixtures. These values were modeled by an artificial neural network (ANN) and the model was tested with root mean squared error (RMSE), mean absolute percentage error (MAPE, %), and regression coefficient (R) values. A higher value of correlation coefficient and smaller values of MAPE and RMSE indicate that the model performs better. For predicting kinematic viscosity at 40 degrees C, correlation coefficients were calculated for training and testing the network as 0.9999 and 0.9995, respectively. Respective MAPE values were determined as 1.011% and 1.8771%. [GRAPHICS] .Öğe Total Organic Carbon Prediction with Artificial Intelligence Techniques(Elsevier Science Bv, 2019) Goz, Eda; Yuceer, Mehmet; Karadurmus, ErdalThis study used the Extreme Learning Machine (ELM), Kernel Extreme Learning Machine (KELM) and Artificial Neural Network (ANN) models with a feed-forward neural network structure and partial least squares (PLSR) methods to estimate total organic carbon. In order to develop models, on-line data measured at five-minute time intervals were collected through one year (2007-2008) from the online-monitoring stations which were built near the River Yesil1rmak in Amasya in North-Eastern Turkey. These stations were the first practice in Turkey. Twelve parameters as luminescent dissolved oxygen (LDO), pH, conductivity, nitrate nitrogen (NO3-N), ammonium nitrogen (NH4-N), total organic carbon (TOC), chloride, orthophosphate, temperature, turbidity, suspended solid and flow rate were measured at the on-line monitoring stations. To predict the total organic carbon, four input variables, pH, conductivity, dissolved oxygen and temperature were selected. Moreover, the data were also collected at the central office in Ankara via a General Packet Radio Service (GPRS) channel. The validity of models was tested by using statistical methods in MATLAB including correlation coefficients (R), mean absolute percentage error (MAPE%) and root mean square error (RMSE). The best result was obtained in the presence of KELM with a radial basis function (RBF) kernel. R-test=0.984, MAPE(test)=3.01, RMSEtest=0.9676. Additionally, R-train=0.995, MAPE(train)=1.58 and RMSEtrain=0.532. Among the other two algorithms ANN provided better results than ELM and PLSR.
