Mühendislik Fakültesi / Faculty of Engineering

Permanent URI for this collectionhttps://hdl.handle.net/11727/1401

Browse

Now showing 1 - 20 of 183

Aintshop Production Line Optimization Using Response Surface Methodology
(2007) Dengiz, Berna; Belgin, Onder; 0000-0001-6702-2608; K-1080-2019
This paper deals with the problem of determining the optimum number of workstations to be used in parallel and workers at some stations using simulation optimization approach in a paint shop line of an automotive factory in Ankara, Turkey. In the optimization stage of the study Response Surface Methodology (RSM) is used to find the optimum levels of considered factors. Simulation model and optimization stage integration is used both to analyse the performance of the current paint shop line and determine the optimum working conditions, respectively, with reduced cost, time and effort.
Analysis of Heart Diseases from ECG Signal
(2014) Kantar, Tugce; Koseoglu, Ovul; Erdamar, Aykut; https://orcid.org/0000-0001-8588-480X; AAA-6844-2019
At the present time, the main method used in the diagnosis of heart disease is the electrocardiography (ECG). The purpose of this study is design a decision support algorithm which automatically detect the normal sinus rhythm or other pathologies. The improved algorithm will provide support to the doctor can be also used for educational purposes. Within the scope of this study, with the design of the rule based algorithm which automatically detect a normal sinus rhythm and non normal sinus rhythm, in total it can detect eight pathologies. In maincode there are thirteen functions that are used for diagnose eight different ECG pathology automatically. Higher success is being anticipated in future for the prediction power of the developed method with continuing research on the matter.
Analysis of Lack of Cohesion in Methods (LCOM): A Case Study
(2021) Haner Kirgil, Elif Nur; Ercelebi Ayyildiz, Tulin
As the need for software has increased, the maintainability of software has become an important issue and the need for the qualitative study of source codes has arisen. For this reason, we used software quality metrics to measure software quality. Moreover, for object-oriented software development projects, cohesion is an important factor. The main purpose of cohesion is to provide the rule that each class should serve a single purpose. Every class should have a single responsibility. In this study, types of Lack of Cohesion in Methods (LCOM) metrics were examined and compared using a sample Java code. As a result, it was observed that the LCOM5 metric gave the best result.
Analysis of True Random Number Generation Method Utilizing FM Radio Signals
(2015) Tanyer, Suleyman Gokhun; Inam, Sitki Cagdas; Atalay, Kumru Didem; 0000-0001-9506-2391; 0000-0003-0820-9186; I-5023-2013
Random number generator (RNG)'s utilize some source of entropy. Various RNGs; pseudo, quasi and true RNGs provide observed samples for given distributions which are necessary in statistical signal processing and simulations applications. True random number generators generally utilize the last insignificant bit of observed samples obtained from a source of entropy. In this work, samples obtained from an FM broadcasting are examined. Least significant bit is used to generate 16 bits observed samples. Randomness and goodness-of-fit test results are compared with the raw samples. It is shown that it is possible to generate true random numbers of sufficient merits using FM signals even when the band is occupied with broadcasting signals.
An Analysis Tool that Detects The Code Caves in Specified Sizes for Portable Executable Files
(2022) Ugurlu, Guney; Acici, Koray; 0000-0002-3821-6419; HDM-9910-2022
Code caves represent sequential null bytes in portable executable files and are particularly used in reverse engineering. With the help of code caves, the execution flow of the program can be changed, and different codes can be injected into the compiled programs. In the sections in the PE files, it is determined manually whether there is a code cave suitable for the size of the code to be injected. This paper presents the analysis tool named CodeCaveFinder. It finds in detail whether the code caves of the user desired size are in the sections of the PE file. As a result of tests, it has been proven that CodeCaveFinder gives accurate and reliable results.
Application of Cellular Manufacturing and Simulation Approaches for Performance Improvement in an Aerospace Company's Manufacturing Activities
(2019) Ic, Yusuf Tansel; Yurdakul, Mustafa; Gulsen, Mehmet; Dalcan, Gizem; Kirdar, Didem; Kale, Meric; Altinkaya, Zeynep; Aktas, Dogukan; https://orcid.org/0000-0001-9274-7467; AGE-3003-2022
This paper presents performance improvement studies performed in manufacturing activities of an aerospace company. Cellular Manufacturing (CM) and simulation approaches are used to improve performance in the factory of the company. In the application of CM, reorganization of machines in the machining section as manufacturing cells is performed by using Rank Order Clustering (ROC) algorithm and Hollier method. The performance improvement provided by CM application is measured in terms of the total material handling distances travelled by the parts. On the other hand, a simulation model in SIMAN is developed for panel production section of the factory. Based on the performance outputs of the simulation model, various recommendations are proposed in the operation rules and numbers of resources in the panel production section. The results show that both approaches provide considerable performance improvement in the manufacturing activities of the factory.
Audio Based Violent Scene Classification Using Ensemble Learning
(2018) Sarman, Sercan; Sert, Mustafa
In this paper, we deal with the problem of violent scene detection. Although visual signal has been widely used in detection of violent scenes from video data, audio modality; on the other hand, has not been explored as much as visual modality of the video data. Also, in some scenarios such as video surveillance, visual modality can be missing or absent due to the environmental conditions. Therefore, we use the audio modality of video data to decide whether a video scene is violent or not. For this purpose, we propose an ensemble learning method to classify video scenes as "violent" or "non-violent". We provide empirical analyses both for different audio features and classifiers. As a result, we obtain best classification performance by using the Random Forest algorithm along with the ZCR feature. We use MediaEval Violent Scene Detection task dataset for the evaluations and obtain superior results with the official metric MAP@100 of 66% compared with the literature.
Audio Captioning Based on Combined Audio and Semantic Embeddings
(2020) Eren, Aysegul Ozkaya; Sert, Mustafa
Audio captioning is a recently proposed task for automatically generating a textual description of a given audio clip. Most existing approaches use the encoder-decoder model without using semantic information. In this study, we propose a bi-directional Gated Recurrent Unit (BiGRU) model based on encoder-decoder architecture using audio and semantic embeddings. To obtain semantic embeddings, we extract subject-verb embeddings using the subjects and verbs from the audio captions. We use a Multilayer Perceptron classifier to predict subject-verb embeddings of test audio clips for the testing stage. Within the aim of extracting audio features, in addition to log Mel energies, we use a pretrained audio neural network (PANN) as a feature extractor which is used for the first time in the audio captioning task to explore the usability of audio embeddings in the audio captioning task. We combine audio embeddings and semantic embeddings to feed the BiGRU-based encoder-decoder model. Following this, we evaluate our model on two audio captioning datasets: Clotho and AudioCaps. Experimental results show that the proposed BiGRU-based deep model significantly outperforms the state of the art results across different evaluation metrics and inclusion of semantic information enhance the captioning performance.
Audio-based Event Detection in Office Live Environments Using Optimized MFCC-SVM Approach
(2015) Kucukbay, Selver Ezgi; Sert, Mustafa; 0000-0002-7056-4245; AAB-8673-2019
Audio data contains several sounds and is an important source for multimedia applications. One of them is unstructured Environmental Sounds (also referred to as audio events) that have noise-like characteristics with flat spectrums. Therefore, in general, recognition methods applied for music and speech data are not appropriate for the Environmental Sounds. In this paper, we propose an MFCC-SVM based approach that exploits the effect of feature representation and learner optimization tasks for efficient recognition of audio events from audio signals. The proposed approach considers efficient representation of MFCC features using different window and hop sizes by changing the number of Mel coefficients in the analyses as well as optimizing the SVM parameters. Moreover, 16 different audio events from the IEEE Audio and Acoustic Signal Processing (AASP) Challenge Dataset, namely alert, clear throat, cough, door slam, drawer, keyboard, keys, knock, laughter, mouse, page turn, pen drop, phone, printer, speech, and switch that are collected from office live environments are utilized in the evaluations. Our empirical evaluations show that, when the results of the proposed methods are chosen for MFFC feature and SVM classifier, the tests conducted through using 5-fold cross validation gives the results of 62%, 58% and 55% for Precision, Recall and F-measure scores, respectively. Extensive experiments on audio-based event detection using the IEEE AASP Challenge dataset show the effectiveness of the proposed approach.
Author Recognition from Lyrics
(2015) Kirmaci, Basar; Ogul, Hasan
Music information retrieval has been an important task due to the wide use of internet and related technologies for entertainment. In previous studies, the problem has been considered using the meta-data or melodic content. The use of lyrics in this context is not that common. There is not study either for Turkish songs in this respect. In this study, we discuss the predictability of the author using the text data in a Turkish lyric. To this end, we propose a system that can predict the author using the features extracted from text content. The performance of the system is evaluated on a large data set collected from writers with different music styles.
Automated Tuberculosis Detection Using Pre-Trained CNN and SVM
(2021) Oltu, Burcu; Guney, Selda; Dengiz, Berna; Agildere, Muhtesem
Tuberculosis (TB) is a dreadfully contagious and life-threatening disease if left untreated. Therefore, early and accurate diagnosis is critical for treatment. Today, invasive, expensive, or time-consuming tests are performed for diagnosis. Unfortunately, accurate TB diagnosis is still a major challenge. In the proposed study, a decision support system that can automatically separate normal and TB chest X-ray (CXR) images is presented for objective and accurate diagnosis. In the presented methodology, first various data augmentation methods were applied to the data set, then pre-trained networks (VGG16, MobileNet), were employed as feature extractors from augmented CXR's. Afterward, the extracted features for all images were fed into a support vector machine classifier. In training process, 5-fold cross-validation was applied. As a result of this classification, it was concluded that TB can be diagnosed with an accuracy of 96,6% and an area under the ROC curve (AUC) of 0,99.
Automatic Brain Tissue Segmentation on TOF MRA Image
(2020) Ozen, Sinasi Kutay; Aksahin, Mehmet Feyzi
For the segmentation of brain vessels from MRA images, brain tissue is used in the head, eye, skull, etc. must be separated from the structures. For this reason, studies are carried out for the segmentation of brain tissue. In this study, the method that automatically segregates brain tissue from magnetic resonance angiography images taken with time of flight (TOF) technique is presented. The method in the study consists of live steps. First of all, the tip contrast values in the image are filtered by anisotropic diffusion filtering method. Parameters of anisotropic diffusion method are determined automatically by the natural image quality evaluator method. Sudden density transitions arc detected by applying LoG edge detection filter on the filtered image. It is made ready for image analysis by applying etching on the image with density transitions. According to the conditions determined in image analysis, brain tissue is obtained separated from other head structures. As a result of this study, an easy-to-apply, fast-delivering, high-accuracy automatic algorithm has been introduced.
Automatic Classification of Respiratory Sounds During Sleep
(2018) Kilic, Erkin; Erdamar, Aykut
Sounds like snoring, coughing, sneezing, whistling, which have different acoustic properties, can emerge involuntarily during the sleep. These sounds may affect negatively the sleep quality of the other people in the same environment, just as it may affect directly the sleep quality. To increase the sleep quality, these sounds should be recorded and evaluated by a sleep expert. This is an expertise required process that can be time-consuming and subjective results. In this study, it has been aimed that developing a computer-aided diagnosing algorithm which will classify the sounds emerging during the sleep automatically with high accuracy by analyzing the records in a fast and effective way to help the sleep expert to diagnose. The mathematical features have been obtained in frequency and time domains by applying continuous wavelet transform for the different type of sounds. Support vector machine used as a classifier. 390 and 449 segments were used for training and testing respectively. As a result of the study, six different parameters which are exhalation, simple snoring, high frequency duplex snoring, low frequency duplex snoring, triplex snoring and coughing were classified with 96.44% accuracy rate.
Automatic Detection of Sleep Spindles With Quadratic Discriminant Analysis
(2018) Kokerer, Sila Turku; Celik, Elif Oyku; Kantar, Tugce; Erdamar, Aykut
Sleep, is in the event of temporary loss of consciousness. Sleep and wakefulness causes some different kind of potential changes in brain. Transient waveforms observed in sleep electroencephalography are structures with specific amplitude and frequency characteristics that can occur in some stages of sleep. The main objective of this study is to develop a method to detect sleep spindle, which is one of these structures, with high-accuracy. Sleep spindles that require the expertise to determine visually, is a process that can be time-consuming and subjective results. In this study, electroencephalography records, scored by expert sleep physicians, were analyzed by different methods. Two features have been determined that express the presence of sleep spindle. Sleep spindles were detected by these features and quadratic discriminant analysis. As a result, the performance of the algorithm was evaluated and sensitivity, specificity and accuracy were determined as 95.74%, 98.08% and 97.76%, respectively.
Automatic Glacoma Detection Using Whale Optimization and Support Vector Machines
(2022) Ozen, Sinasi Kutay; Aksahin, Mehmet Feyzi
Glaucoma is among the most common causes of permanent blindness in humans. The mass screening will aid in early diagnosis in a large population, as the initial symptoms are not obvious. This type of mass screening requires an automated diagnostic technique. Our proposed automation extracts feature by obtaining disk-to-cup ratio by applying histogram equalization, median filter, otsu thresholding, and whale optimization algorithm, respectively, on the optic disc region obtained by preprocessing. In addition, the optic disc circumference, optic disc area, optic cup circumference, and optic cup area values obtained from the optic disc region are given to the support vector machine model together with the cup-disc ratio, and glaucoma detection is made automatically. The proposed system has been validated on a real ophthalmological images of both normal and glaucoma cases. The results show the effectiveness of the proposed method when compared with other existing systems.
Automatic Vascular Segmentation on Angio Images
(2017) Akshain, Mehmet Feyzi; Ozen, S. Kutay; Eren, Neyyir Tuncay
Cardiovascular disease is one of today's major health problems. These diseases are the result of constriction or blockage of coronary vessels feeding heart. Diagnosis of Cardiovascular contraction is determined visually by physicians with angio imaging method. Visually defined vascular diseases may give subjective results. Vessel constrictions in angiograms are automatically determined by vascular segmentation to help physicians diagnose cardiovascular stenosis and to minimize subjective results. In present study, adaptive thresholding method and frangi filter were applied to the pre-processed angiogram images. After this application, the noise cleaning method on the determined cardiovascular image was performed and the vein structure was detected with high accuracy rate and low calculation time.
Beam Steering of Vortex Waves by A Phased Array Based on The Field Equivalence Principle
(2022) Hizal, Altunkan; Yildiz, Hayrullah
A further ability can be added to an existing phased array (PA) to include generation and steering of noncollimated or collimated electromagnetic (EM) vortex waves (VW). The concept is based on the EM field equivalence principle. The near field of VW's generated by a uniform circular array (UCA) is calculated on a tilted planar finite size reference aperture (RAP) which intercepts all the VW modes. Using the Love equivalence principle and the fields of UCA on the RAP, the VW's are calculated in the far field. RAP is divided into small rectangular subapertures (SAP), simulating the elements of a PA. The UCA-fields on each SAP, for a given RAP's tilt angle (steering angle) are calculated. These fields are to be generated by rightly polarized PA's antenna elements fed by the associated transmit-receive modules. The method is also applied to VW's tightly collimated by a paraboloidal reflector. Numerical simulations obtained verified this concept.
A Case Study on the Utilization of Problem and Solution Domain Measures for Software Size Estimation
(2016) Ayyildiz, Tulin Erelebi; Kocyigit, Altan; 0000-0002-7372-0223; 0000-0001-5003-4127; AAE-1726-2021; S-6347-2016
Detailed requirements is the primary input of software size measurement and effort estimation methodologies and a significant amount of time and expertise is needed for size measurement. In order to streamline size measurement and effort estimation, this study exploits the correlations between the problem domain measures such as the number of distinct nouns and distinct verbs in the requirements artifacts and the solution domain measures such as the number of software classes and methods in the corresponding object oriented software. In this case study, 12 commercial software projects are analyzed and multiple regression analysis is carried out to develop an estimation model for the solution domain metrics in terms of problem domain metrics. The results suggest that, for the projects examined, it is possible to use problem domain measures to make plausible predictions for the solution domain metrics.
CFD vs. XFOIL of Airfoil Analysis at Low Reynolds Numbers
(2016) Gunel, Onur; Koc, Emre; Yavuz, Tahir
For Blade Element Momentum (BEM) theory, the airfoil data needs to be as accurate as possible. Nowadays, Computational Fluid Dynamics (CFD) is used for optimization and design of turbine application. Lift coefficient, drag coefficient and lift coefficient over drag coefficient are significant parameters for turbine application. Selecting a suitable computational tool is crucial for design of the turbine. Panel method and an integral boundary layer formulation are combined in the MOIL code for the analysis of potential flow around the airfoils. In this study, XFOIL code and Transition SST k-omega model were used to predict the aerodynamic performance at low Reynolds numbers (Re = 3x10<^>5, 4x10<^>5, 5x10<^>5). The results were compared and CFD results and XFOIL, code result are compatible with each other until stall angle. Also XFOIL, and CFD results were shown a good coherence with literature.
Change Detection of Urban Areas in Ankara through Google Earth Engine
(2018) Celik, Naime
Rapid urbanization with inadequate planning can have negative impact on the cities with growing population. Maintaining sustainable urban development and planning becomes one of the major necessity for the government and policy makers. Thus, being able to monitor the changes and the enlargement of the cities provides a valuable information for those decision-making bodies. This study investigates the possibilities of identifying the changed areas with Google Earth Engine (GEE) providing a fast and easy-to-use platform with its geospatial analysis tools for applications such as change detection. In this study, C-band Synthetic Aperture Radar (SAR) images of Sentinel-1 and Multispectral Instrument (MSI) images of Sentinel-2 were utilized to identify the changed areas, such as new built-ups or soil excavation, of 21 neighborhood units of capital city of Turkey through GEE platform. The image subtraction of MSI images were integrated with image subtraction of SAR images of 2015 and 2017. The image indices such as Difference Built-up Index (NDBI), Bare Soil Index (BSI) and Soil-adjusted Vegetation Index (SAVI) were also utilized. A binary supervised classification was performed by using Random Forest classifier. Finally, a post-processing with morphological operators was conducted to reduce the effects of pixel-based classification and to achieve higher test accuracy. With the post-processing, 91% overall test accuracy and kappa value of 0.82 were achieved. The study reveals an average of 5.9% change of the total area in those selected 21 neighborhood units of Ankara and Erler neighborhood unit is the most altered with 14.5% of its total area.