İnşaat Mühendisliği Bölümü Koleksiyonu
Bu koleksiyon için kalıcı URI
Güncel Gönderiler
Öğe Application of state-of-the-art machine learning algorithms for slope stability prediction by handling outliers of the dataset(Springer, 2023) Demir, Selçuk; Şahin, Emrehan KutluğThis paper addresses the issue of the prediction of slope stability with machine learning (ML) applications. Five well-known and popular ML algorithms, namely neural network (NNet), decision tree (DT), support vector machine (SVM), k-nearest neighbor (kNN), and random forest (RF), are used to demonstrate the effectiveness of the ML algorithms for predicting binary classification of slope stability based on a case history dataset containing outliers. This study also evaluates the winsorization method used to treat outliers in the dataset by outlining the effect of outliers on the prediction performances of models. To this end, the performance of all the generated ML models is assessed and compared both for unwinsorized (e.g., raw) and winsorized datasets based on performance metrics (i.e., Recall, Precision, Accuracy, and F1-Score) obtained from the confusion matrix. The experimental outputs showed that the application of winsorization enhanced the prediction performance of the models, and thus, all ML models built with winsorized datasets outperformed the unwinsorized ones. In this paper, the RF model achieves the best prediction performance, especially in the case of the winsorized dataset used. Moreover, it is found that SVM is the most sensitive algorithm to outliers as against the other ML algorithms, while the kNN algorithm is the least among the applied algorithms. Results showed that the increment percentage of accuracy nearly reaches 20% for the SVM model and the following 18% for DT, 11% for NNet, 10% for RF, and 4% for kNN, respectively. Furthermore, the results of the study reveal not only the performance of ML algorithms for the slope stability problem but also show how the handling of outliers of a dataset affects the models' prediction performance.Öğe Assessing the predictive capability of DeepBoost machine learning algorithm powered by hyperparameter tuning methods for slope stability prediction(Springer, 2023) Demir, Selçuk; Şahin, Emrehan KutluğThis paper presents DeepBoost based classification model for the slope stability problem, wherein an extensive dataset consisting of six features is used. The developed DeepBoost model is trained and tested with 444 stable and unstable slope cases. For comparison, the predictive performance of DeepBoost is systematically compared with the other state-of-the-art ML algorithms, i.e., Adaptive Boosting (AdaBoost.M1) and Support Vector Machine (SVM) based on the well-established confusion matrix, which contains the known metrics of Accuracy (Acc), Precision (P), Recall (R), F1-Score (F), and Kappa Score (kappa). Furthermore, three hyperparameter optimization approaches, Grid Search (GS), Random Search (RS), and Bayesian Optimization (BO), have been integrated for tuning the hyperparameters of the DeepBoost and the other models to achieve the best results. Based on the comparative analysis, it was found that BO optimized DeepBoost model achieved the best performance score and accurately detected and classified all types of slope stability scenarios. Also, Bayesian optimized models performed better than GS and RS optimized ones. As a result, the comparison results of the developed DeepBoost model with the other models reveal that DeepBoost exhibited superior performance as compared to the other algorithms in the case of BO with an accuracy of Acc = 96.97% for DeepBoost, Acc = 95.45% for AdaBoostM1, and Acc = 90.91% for SVM.Öğe Maksimum ve kalıcı yer değiştirme talepleri açısından TBDY 2018 ve DBYBHY 2007 deprem yönetmeliklerinin karşılaştırılması(Pamukkale University, 2023) Demir, AhmetSeismic codes are updated by taking into account the information obtained as a result of scientific studies and the observations after earthquakes about the behavior of the building. In Turkey, the Turkish Building Earthquake Codes (TBEC) has also been published in 2019, instead of the Turkish Earthquake Codes (TEC). Both seismic codes contain definitions for nonlinear dynamic analysis for the design and/or evaluation of structures. However, in TBEC, both the definition of the design spectrum and the spectral parameters used to obtain the design spectrum for any location in Turkey have changed. In order to evaluate the effects of these changes on the drift demands, nonlinear dynamic analyzes of single degree of freedom (SDOF) systems with different periods and lateral strength ratios were made according to both TBEC and TEC. Then, mean of the maximum and residual drift demands and the scattering of these demands were compared. For this purpose, totally 72 different TSD systems were considered. For the analysis of these systems, 84 and 168 ground motion record sets were used according to TEC and TBEC (different earthquake zones, soil classes and cities), respectively. When the results are examined; it is seen that a) the design spectrums change according to the soil classes and cities, b) the maximum and residual drift demands change depending on the earthquake levels and soil classes, c) in some cities, the demands obtained with TBEC, and in some cities, the demands obtained with TBEC are higher, d) drift demands scattering within the set is high.Öğe Evolution of operable slip systems, lattice strain fields and morphological view of Bi-2223 ceramic system with optimum NiO addition(Elsevier, 2023) Mercan, Ali; Kara, Emre; Doğan, Muhsin Uğur; Kaya, Şenol; Terzioğlu, Rıfkı; Erdem, Ümit; Yıldırım, Gürcan; Terzioğlu, CabirThe current work extensively reveals the influence of different nickel oxide (NiO) impurity addition levels on the morphological, microstructural, key mechanical performance, and mechanical characteristic properties of Bi1.8Pb0.4Ca2.2Sr2Cu3Oy (Bi-2223) ceramics using scanning electron microscopy (SEM), powder X-ray diffraction (XRD), and Vickers micro-indentation (Hv) hardness measurements. It was observed that the addition of NiO impurity in the Bi-2223 crystal structure affected seriously the fundamental characteristic features. In the case of the optimum NiO concentration level of x = 0.1, the Bi-2223 materials exhibited the best crystallinity quality and coupling strengths between the adjacent layers, the most uniform surface view, and the densest, and the smoothest crystal structure. Similarly, the compound was noted to possess the hardest, highest mechanical strength, durable tetragonal phase, resistance toward failure by fatigue, and elastic recovery properties. Besides, it was observed that the characteristic Bi-2223 superconducting phase fraction and stabilization of the tetragonal crystal system reached the maximum level for the optimum concentration. Moreover, optimum NiO particles brought about a considerable increase in the number of operable slip systems, surface residual compressive force regions, and lattice strain fields. Correspondingly, the mobility of defects was blocked significantly depending on the preference of defects through transcrystalline regions. Additionally, optimum addition strengthened the typical indentation size effect due to the improvement of the recovery mechanism. In this regard, the NiO-added sample exhibited the least response to the applied loads. Thus, the Bi-2223 sample with the optimum NiO concentration was found to present the highest hardness parameter of 0.496 GPa, greatest elastic deformation value of 16.493 GPa, largest stiffness value of 1.044 MN/m, and smallest contact depth of 5.849 mu m. On the other hand, after the optimum concentration level of x = 0.1, there appeared serious increase in problems including internal defects, impurity residues, microscopic structural problems, and connection problems between the grains. All experimental findings were theoretically supported by semi-empirical mechanical methods. To sum up, the addition of NiO particles was noticed to increase the potential application areas of Bi-2223 ceramicÖğe An environmentally friendly approach to soil improvement with by-product of the manufacture of iron(North Carolina State Univ Dept Wood & Paper Sci, 2023) Keskin, İnan; Şentürk, İbrahim; Yumrutaş, Halil İbrahim; Totiç, Ermedin; Ateş, AliBlast furnace slag has been used for many years in various applications related to civil engineering. Many studies have created a wide variety of cost-effective and environmentally friendly solutions for this industrial byproduct. This study aims to contribute to the performance evaluations of the usability of the blast furnace slag for soil improvement and the effects of the additive ratio and curing time. Bentonite samples were prepared with the addition of blast furnace slag at 5%, 10%, 15%, and 20% ratios by weight at optimum water content (wopt). Results were evaluated using the liquid limit, plastic limit, unconfined compressive strength, and swelling tests performed after 1, 7, 14, and 28 days of curing time. Results revealed that the liquid limit value decreased, and the unconfined compressive strength increased with increasing curing time and blast fumace slag ratio in the mixture. Additionally, swelling pressure generally decreased with increasing slag contribution and curing time. The lowest values of the unconfined compressive strength were observed on the 7th day of curing time, and the minimum value was obtained at 10% mixing ratio. The highest unconfined compressive strength values were observed on the 28th days of curing time. The optimum mixing ratio was 5%.Öğe Improvement in deformation degree of Zr surface-layered Bi-2223 ceramics by diffusion annealing temperature(Elsevier Sci LTD, 2023) Mercan, Ali; Terzioğlu, Rıfkı; Doğan, Muhsin Uğur; Kaya, Şenol; Erdem, Ümit; Yıldırım, Gürcan; Terzioğlu, Cabir; Varilci, AhmetThis study investigated the effects of different annealing temperatures (650 degrees C <= T <= 840 degrees C) on the surface morphological and mechanical performance properties of Zr surface-layered Bi-2223 materials with scanning electron microscopy (SEM) images, Vickers microhardness (Hv) measurements, and semi-empirical mechanical approaches. It was observed that the ceramic compound exposed to 650 degrees C annealing temperature exhibited the superior performance features due to the enhancement in the deformation degree. This is because the Zr ions behaved as the nucleation centers to prevent the propagations of cracks and dislocations throughout the main matrix depending on the decrease in the degree of granularity and distributions of crystal structure problems over a wider area. Similarly, the SEM pictures indicated that the diffusion mechanism increased the random distributions of the thinner plate-like granular structures (serving as nucleation centers), leading the decrease in the coupling problems between the grains. Among the materials, the highest surface densification was observed for the compound exposed to 650 degrees C. Namely, surface morphological analysis showed a strong correlation be-tween microstructure and mechanical performances. Further, the zirconium ions were found to decrease in the non-recoverable stress concentration sites, crack-initiating defects, and dislocations in the ceramic system. Accordingly, the sensitivity to the applied test load was noted to decrease dramatically. Shortly, crack growth size and velocity were observed to be more easily under control. Correspondingly, the Zr ions delayed consid-erably the beginning points of saturation limit (load-independent) regions for the bulk Bi-2223 superconducting materials. Additionally, the Zr ions led to the change in the mechanical characteristic behavior from typical indentation size effect to reverse indentation size effect. Lastly, the microindentation hardness measurements were semi-empirically analyzed by the different models. According to the comparison, Hays-Kendall mechanical model was noted to provide the closest parameters to the load-independent microhardness results.Öğe Multi-functional solution model for spectrum compatible ground motion record selection using stochastic harmony search algorithm(Springer, 2022) Kayhan, Ali Haydar; Demir, Ahmet; Palancı, MehmetSelection of appropriate ground motion (GM) records for nonlinear dynamic analysis (NDA) of structures plays a crucial role to estimate structural responses reasonably. In this study, a multi-functional solution model utilizing stochastic harmony search (HS) algorithm is proposed to obtain scaled or unscaled real GM component sets for uni-directional analysis of two-dimensional structural models and GM component pair sets for bi-directional analysis of three-dimensional structural models. The solution model allows to consider compatibility between target spectrum and both mean spectrum and individual spectra besides desired spectral variability. Uniform hazard spectrum, conditional mean spectrum or scenario-based spectrum can be selected as target spectrum. Combined response spectra of selected component pairs such as SRSS, geometric mean and maximum directional can also be handled by the solution model. To demonstrate the efficiency of the solution model, various examples were presented. In addition, a sensitivity analysis was performed to evaluate the effect of HS parameters on the solution accuracy. Results show that the proposed solution model can be regarded as efficient to obtain appropriate GM record sets to be used for NDAs within a probabilistic seismic design and/or performance assessment framework.Öğe Investigation of the effect of real ground motion record number on seismic response of regular and vertically irregular RC frames(Elsevier Science Inc, 2022) Demir, AhmetNonlinear time history analysis is an analytical method generally used in performance-based seismic design. With this method, seismic responses are obtained more realistically. Selection of ground motion records for nonlinear time history analysis is an important step since it strongly affects the analysis results. Therefore, it has always been a matter of curiosity to investigate the effect of the characteristics, content and number of the records on the analysis results. In this study, seismic responses of regular and irregular RC frames were investigated by varying the number of real ground motion records in a set. For this purpose, 13 different groups that contain three to hundred real ground motion records in size in a set have been considered and ten different earthquake record sets are obtained for each group. Ground motion selection procedure of Eurocode-8 was considered and a total of 130 sets were used for nonlinear response analysis of RC frames. Global drift ratio, maximum floor acceleration, inter-story drift ratio and six different intensity measures (IMs) were used to investigate the effect of the number of records. According to analysis results, nonlinear responses of RC frames are more stable and might be sufficient when the number of real records in a set is higher than seven according to Eurocode-8. Results indicate that if the number of real records in a set are lower than seven, conservative seismic responses can be found since the maximum rather than mean response values are used. It is observed that dispersion of seismic demands and mean to median ratios can be increased if the number of real records in a set is higher than ten. In addition, the correlation between some of IMs and seismic demands increase when the number of records in a set increased from three to seven and it remains stable from seven to hundred records. Furthermore, 7, 8, 15 and 22 records show the lowest error terms of considered engineering demand parameters for regular and irregular RC frames.Öğe Liquefaction prediction with robust machine learning algorithms (SVM, RF, and XGBoost) supported by genetic algorithm-based feature selection and parameter optimization from the perspective of data processing(Liquefaction prediction with robust machine learning algorithms (SVM, RF, and XGBoost) supported by genetic algorithm-based feature selection and parameter optimization from the perspective of data processing, 2022) Demir, Selçuk; Şahin, Emrehan KutluğLiquefaction prediction is an important issue in the seismic design of engineering structures, and research on this topic has been continuing in current literature using different methods, including experimental, numerical, or soft computing. In this paper, three robust machine learning (ML) algorithms are applied to predict soil liquefaction using a set of 411 shear wave velocity case records. The Genetic Algorithm (GA) based feature selection (FS) and parameter optimization of Random Forest (RF), Support Vector Machines (SVM), and eXtreme Gradient Boosting (XGBoost) algorithms are utilized to improve the accuracy of the liquefaction prediction models. Simple Random Sampling (SRS) and Stratified Random Sampling (StrRS) are used for data sampling, and also SMOTE algorithm are applied to prepare the balanced training sets. The results of robust ML algorithms are assessed based on well-known five performance matrices, namely Accuracy (Acc), Kappa, Precision, Recall, and F-Measure. Evaluation of the results is made separately for each ML algorithm considering sampling data generated from SRS, StrRS, and SMOTE. As a result, the XGBoost model is more accurate (Acc = 96%) than RF (Acc = 93%) and SVM (Acc = 91%) in the case of the SMOTE algorithm. This study reveals the superiority of the XGBoost algorithm in the liquefaction prediction and shows how the accuracy measures tend to improve when the predictive models are trained using balanced samples with StrRS and SMOTE sampling strategies.Öğe Comparison of tree-based machine learning algorithms for predicting liquefaction potential using canonical correlation forest, rotation forest, and random forest based on CPT data(Elsevier Science Ltd, 2022) Demir, Selçuk; Şahin, Emrehan KutluğThis research investigates and compares the performance of three tree-based Machine Learning (ML) methods, Canonical Correlation Forest (CCF), Rotation Forest (RotFor), and Random Forest (RF), for predicting the liquefaction potential of soils based on the cone penetration test (CPT) case history datasets collected from previously published research. The ML models are trained and validated using the Stratified Random Sampling technique for training and test datasets considering three sampling ratios as 50:50, 40:60, and 70:30. In addition, a comparative example was applied to show the difference between the Stratified Random Sampling and the Simple Random Sampling technique, which is the most common probability-based sampling method, considering only a dataset. The predictive capabilities of the developed models are evaluated using Overall Accuracy, Kappa, Precision, Recall, and F-Measure values. Lastly, the Wilcoxon Signed-Rank Test and the Pearson's Correlation Coefficient are adopted to determine the statistical significance of the accuracies between the tree-based ML methods. Generally, tree-based ML methods of CCF, RotFor and, RF are found robust with respect to the variations in training sample sizes, and the performance metrics revealed that the CCF and RotFor method exhibited slightly better performance than the conventional RF method. Finally, based on the results obtained from performance assessment output, CCF and RotFor methods which are the first application in the soil liquefaction issue to the best of our knowledge are worth considering in the prediction of soil liquefaction.Öğe Numerical investigation of the effects of ground motion characteristics on the seismic behavior of liquefiable soil(Budapest University of Technology and Economics, 2023) Demir, SelçukThe seismic behavior of liquefiable soils can be significantly influenced by many ground motion characteristics. Therefore, it is crucial to identify the ground motion characteristics that have the most significant effects on the seismic behavior of liquefiable soils. In this paper, a series of nonlinear numerical analyses were performed to investigate the influence of ground motion characteristics on the seismic behavior of loose liquefiable soil. The liquefiable soil profiles were built with the same relative densities but different layer thicknesses. In order to clarify the effect of the ground motion characteristics on the liquefiable soil mechanism, soil profiles were subjected to ground motion sets having different characteristics, such as maximum horizontal accelerations, frequency contents, and significant durations. The numerical analyses were performed using the open-source program OpenSees. The results were presented and discussed in terms of peak ground acceleration, amplification ratio, maximum excess pore pressure ratio, maximum shear strain, and maximum lateral displacement. The results indicated that the maximum horizontal acceleration and the frequency content greatly influence the site response behavior of the liquefiable soil. Furthermore, the nonlinear behavior of the soil is more obvious on being subjected to long-duration ground motions as compared to shorter duration ground motions having the same maximum horizontal acceleration. The findings presented in this study could be helpful when analyzing the seismic response of liquefiable soils coupling superstructures.Öğe Greedy-AutoML: A novel greedy-based stacking ensemble learning framework for assessing soil liquefaction potential(Pergamon-Elsevier Science Ltd, 2023) Şahin, Emrehan Kutluğ; Demir, SelçukAutomated machine learning (AutoML) is a generic term for a specific approach to machine learning (ML) area that tries to automate the end-to-end process of employing repetitive ML tasks for real-world problems. In recent years, the AutoML framework, which is the subject of an increasing number of research articles, has become a potential approach for developing complicated ML models without human experience and support. Although existing techniques on AutoML have yielded promising results, research in this field is immature, and new approaches should be developed gradually. This study describes a novel AutoML framework to predict soil liquefaction potential problem based on stacking ensemble learning (SEL) combined with a greedy search algorithm. A special AutoML framework, called Greedy-AutoML, is presented that automatically produces an optimized ML model for predicting on a supervised classification task. The general concept of the proposed AutoML framework consists of three main steps: data preparation, greedy feature selection, and greedy stacking ensemble. Furthermore, the Greedy-AutoML framework has been published on a user-friendly web-based platform for testing or trial purposes. To highlight the capability of this AutoML application, Greedy-AutoML is applied to predict the liquefaction potential of soils using three well-known datasets (i.e., CPT - cone penetration test, SPT - standard penetration test, and Vs - shear wave velocity test) collected from previously published research. The results are assessed based on different performance matrices, namely Accuracy (Acc), Kappa, Precision, Recall, and F1-Score. Experiments with datasets from existing case histories with varying distribution and features showed that the proposed greedy based SEL method achieved an Acc of 98% for the CPT and SPT datasets, while the achieved Acc for the Vs dataset was about 99%.Öğe An investigation of feature selection methods for soil liquefaction prediction based on tree-based ensemble algorithms using AdaBoost, gradient boosting, and XGBoost(Springer London Ltd, 2023) Demir, Selçuk; Şahin, Emrehan KutluğPrevious major earthquake events have revealed that soils susceptible to liquefaction are one of the factors causing significant damages to the structures. Therefore, accurate prediction of the liquefaction phenomenon is an important task in earthquake engineering. Over the past decade, several researchers have been extensively applied machine learning (ML) methods to predict soil liquefaction. This paper presents the prediction of soil liquefaction from the SPT dataset by using relatively new and robust tree-based ensemble algorithms, namely Adaptive Boosting, Gradient Boosting Machine, and eXtreme Gradient Boosting (XGBoost). The innovation points introduced in this paper are presented briefly as follows. Firstly, Stratified Random Sampling was utilized to ensure equalized sampling between each class selection. Secondly, feature selection methods such as Recursive Feature Elimination, Boruta, and Stepwise Regression were applied to develop models with a high degree of accuracy and minimal complexity by selecting the variables with significant predictive features. Thirdly, the performance of ML algorithms with feature selection methods was compared in terms of four performance metrics, Overall Accuracy, Precision, Recall, and F-measure to select the best model. Lastly, the best predictive model was determined using a statistical significance test called Wilcoxon's sign rank test. Furthermore, computational cost analyses of the tree-based ensemble algorithms were performed based on parallel and non-parallel processing. The results of the study suggest that all developed tree-based ensemble models could reliably estimate soil liquefaction. In conclusion, according to both validation and statistical results, the XGBoost with the Boruta model achieved the most stable and better prediction performance than the other models in all considered cases.Öğe Examining the role of liquefiable layer thickness and depth on the seismic lateral response of piles through numerical analyses(ASCE-AMER SOC CIVIL ENGINEERS, 2023) Ari, Abdulmuttalip; Demir, Selçuk; Özener, PelinIn nature, soil layers possess large variability during their geological process. This variability may also lead to differences in the location and thickness of nonliquefiable and liquefiable soil layers. In practice, the impact of liquefaction on the pile can be ignored at depths greater than 20 m due to high confining stress levels and a lack of liquefaction triggering data. On the other hand, this approach may underestimate design loads in many cases, especially in deep-seated and embedded engineering structures. This paper presents the bending response of piles installed through liquefiable layers located at depths beyond 20 m, and parametric analysis was conducted for a wide range of liquefiable layer thicknesses and depths by using OpenSeesPL (version 3.0.2) software. The numerical results were evaluated considering the inelastic concrete pile behavior under different earthquake records and different peak ground accelerations. The findings show that liquefaction can lead to a failure of piles even at depths greater than 20 m, and thus, a design consideration of piles may require a more comprehensive view considering the liquefiable layer depth and thickness effect.Öğe Effect of shear strain compatibility and incompatibility approaches in the design of high modulus columns against liquefaction: A case study in Christchurch, New Zealand(Springer, 2022) Demir, Selçuk; Özener, PelinNowadays, investigating the effectiveness of high modulus columns in liquefaction mitigation is one of the important tasks in earthquake geotechnical engineering. Although there is limited data from the field and laboratory to verify the performance of high modulus columns (HMCs), available case histories, physical model tests, and reliable numerical methods provide important information in order to analyze the role of HMCs in liquefaction mitigation. In this paper, the seismic performance of a liquefied site improved with rammed aggregate piers (RAPs) is investigated through the results of a full-scale field test. Results of cone penetration test (CPT) and cross-hole shear wave velocity (Vs ) test before and after RAP treatment at the test site are assessed to achieve properties of the natural (unimproved) soil, RAP, and the surrounding (improved) soil. The effectiveness of RAPs in liquefaction mitigation is evaluated in terms of pre-and post-improvement factor of safeties against liquefaction, liquefaction-induced deformations, and ground failure indices, which are calculated using shear strain compatibility and incompatibility approaches. The research results showed that RAPs exhibit a satisfying performance when computations are made considering shear strain compatibility in the computation of seismic shear stress reduction factor. On the contrary, the effectiveness of RAPs during the shear strain incompatibility approach is significantly smaller than the ones computed from the current design method based on shear strain compatibility approach. The findings of this study provide a basis for the performance-based ground improvement design through HMCs to mitigate soil liquefaction and also extend knowledge about HMC-improved seismic soil response by presenting the results of liquefaction vulnerability parameters before and after soil improvement of a field test study.Öğe Predicting occurrence of liquefaction-induced lateral spreading using gradient boosting algorithms integrated with particle swarm optimization: PSO-XGBoost, PSO-LightGBM, and PSO-CatBoost(SPRINGER HEIDELBERG, 2023) Demir, Selçuk; Şahin, Emrehan KutluğLiquefaction-induced lateral spreading that has resulted in devastating damages to lifelines and buildings has been widely reported in recent earthquakes. Although it is impossible to preclude the occurrence of earthquakes, it is possible to predict its adverse effects through computer science such as machine learning (ML) algorithms. In this study, the ability of recently developed and powerful ML algorithms such as eXtreme Gradient Boosting (XGBoost), Categorical Boosting (CatBoost), and Light Gradient Boosting Machine (LightGBM) was investigated to predict the occurrence of liquefaction-induced lateral spreading. A relatively large dataset was used to develop ML models, including 6704 lateral spread observations from the 2011 Christchurch earthquake in New Zealand. The particle swarm optimization (PSO) algorithm is utilized for hyperparameter optimization of the gradient boosting models, called the PSO-XGBoost, PSO-LightGBM, and PSO-CatBoost. For comparison, the prediction results of the PSO optimized gradient boosting models were compared with that of the models using default parameters (i.e., XGBoost, CatBoost, and LightGBM). In addition, the SHapley Additive exPlanations approach is employed to explore the feature importance of the variables included in the dataset. The findings demonstrated that all the three gradient boosting algorithms performed well in predicting lateral spreading occurrence. Moreover, PSO-CatBoost outperformed other state-of-the-art models in terms of performance metrics. However, the PSO-LightGBM model may be considered the best choice for computers with older-gen hardware and important tasks that need to be completed in a short time. This study confirms the effectiveness of the proposed models, and the use of these boosting algorithms especially optimized with PSO is recommended for predicting the occurrence of liquefaction-induced lateral spreading.Öğe Implementation of free and open-source semi-automatic feature engineering tool in landslide susceptibility mapping using the machine-learning algorithms RF, SVM, and XGBoost(SPRINGER, 2023) Şahin, Emrehan KutluğVarious machine learning (ML) techniques have been recommended and used in the literature to produce landslide susceptibility map (LSM). On the other hand, feature engineering (FE) is an important topic in ML studies, but the concept is ignored by most research. In this study, a novel FE framework, including feature selection, feature transformation, feature binning, and feature weighting, is proposed to produce LSMs using eXtreme gradient boosting (XGBoost), random forest (RF), and support vector machine (SVM). For this purpose, first, thirteen landslide conditioning factors used in data preprocessing were utilized for producing LSM models in the study area, Babadag district of Denizli Province in the Aegean region of Turkey. Second, two irrelevant factors eliminated from the input feature subset using the feature selection in the FE framework. Third, features determined as skewed data were converted into symmetric form by applying feature transformation analysis with log transformation. Then, the remaining factors having continuous values were turned into categorical values using the quantile classifier technique. During the feature weighting phase, four different feature weighting methods, namely, eXtreme Gradient Boosting, random forest (RF), non-negative least squares (NNLS), and Frequency Ratio, were utilized to calculate the weights in each subclass of each landslide-related factor. In addition, the proposed feature subsets were also compared with raw data. At the end of process, the XGBoost model constructed with a FR-selected subset (Overall Accuracy (Acc) = 0.907 and area under curve (AUC) = 0.9822) outperformed both raw (Acc = 0.874; AUC = 0.960) and other methods (i.e., RF-FR and SVM-NNLS). Consequently, the study results revealed that the proposed FE approach could be a useful framework to increase the performance of ML techniques in identifying and extracting relevant features to develop highly optimized and enriched models.Öğe A comprehensive mechanical response and dynamic stability analysis of elastically restrained bi-directional functionally graded porous microbeams in the thermal environment via mixed finite elements(Springer Heidelberg, 2022) Mollamahmutoğlu, Çağrı; Mercan, Ali; Levent, AykutWith the development of material science and technology, the usage area of microstructures has increased. Microbeams, which are the most common components of these, have attracted the attention of many researchers. A lot of research has been done on free vibration, buckling and dynamic stability behavior for microbeams. In this study, Mixed Finite Element Method (MFEM) is used. Here, free vibration, buckling and dynamic stability analysis are performed for the Timoshenko type elastically restrained bi-directional functionally graded porous microbeams in the thermal environment. In the calculation procedure, microscale effects are based on Modified Couple Stress Theory and Hamilton Principle is used to obtain the governing equations. A functional including field equations and boundary conditions (BCs) are obtained by using the Gateaux differential. One of the important advantages of MFEM is the use of C-0 type shape functions. Another major advantage is that shear locking is not observed with this method. In addition, the formulation allows closed-form integration through the sparse structure of element matrix. The reliability and accuracy of MFEM were demonstrated by comparing the results with the results of the Differential Quadrature Method from the literature for different BCs. In this work, the effects of Winkler and Pasternak elastic foundation parameters (K-w and K-p), AFG and FG power indexes (Px and Pz), thermal environment (Delta T) and partially or fully porosity (ss) were investigated. Finally, mechanical response due to the microscale structure of the problem can be directly obtained without additional processing. Examples related with inhomogeneities related with porosity were given as versatility as well.Öğe Quantifying the effect of amplitude scaling of real ground motions based on structural responses of vertically irregular and regular RC frames(Elsevier Science Inc, 2023) Palanci, Mehmet; Demir, Ahmet; Kayhan, Ali HaydarIncreasing number of real ground motion (GM) record databases raised nonlinear dynamic analysis (NDA) as an attractive option to determine structural response statistics. Structural responses are sensitive to input motions and the appropriate selection and scaling of real GMs is one of the crucial topics in earthquake engineering. In this paper, the influence of amplitude scaling on important earthquake demand parameters (EDPs), namely, global drift ratio, inter-story drift ratio and maximum floor acceleration, were studied using different scaling approaches and scaling limits. Eurocode-8 was used as a reference seismic code and amplitude scaling effects on structural responses of vertically irregular and regular structures were quantified. Efficiency and sufficiency of amplitude scaling were assessed in terms of mean, dispersion and non-exceedance probability curves of the EDPs. Statistical distribution of GM characteristics and their dependence on GM amplitude scaling were also discussed. Evaluations have shown that similar mean responses can be obtained regardless of scaling limits, approaches, and building topology if spectral shape compatibility is ensured. Furthermore, results demonstrated that neither building regularity nor scaling of GMs influenced the statistical distribution of ground motion parameters and non-exceedance probability curves of the EDPs. In fact, it was revealed that record selection scenario including spectral compatibility of individual GMs had a dramatic impact on dispersion and exceedance probability of structural responses.Öğe Probabilistic assessment for spectrally matched real ground motion records on distinct soil profiles by simulation of SDOF systems(Techno-Press, 2021) Demir, Ahmet; Palancı, Mehmet; Kayhan, Ali HaydarSelection of appropriate ground motion records for dynamic analysis has uttermost importance since it significantly affects structural responses which are used for seismic performance assessment of buildings. This study focuses on probabilistic assessment of several record selection strategies that apply different level of constraints for spectrally matched real ground motion records. For this purpose, single degree of freedom (SDOF) systems with various lateral strength capacity ratios, vibration periods and hysteretic models were considered to cover broad type of structural systems and maximum displacement demands of SDOF systems were obtained by nonlinear dynamic analyses. Using the analysis results, central tendency of maximum displacement demands was evaluated. Confidence intervals of the demands were also estimated in probabilistic manner. In addition, non-exceedance probability curves of the displacement demands were constructed. Results indicate that using supplementary constraints about spectral matching, it is possible to control the variation of spectral accelerations and hence the variation of seismic displacement demands. In conclusion, displacement demands can be obtained for code-or probability-based design/performance assessment with appropriate selection approach considering desired variation which can be determined from either probabilistic or deterministic seismic hazard analysis.
