Murat Selek

Using artificial neural networks for real-time observation of the endurance state of a steel specimen under loading

The surface temperature behavior of a steel specimen under bending fatigue is exactly divided into three stages: an initial temperature increase stage, a constant temperature stage and an abrupt temperature increase stage at the end of which the specimen fails. To obtain the endurance state of the specimen we use its thermal images (TIs). By applying artificial neural networks (ANNs) and other operations to these TIs we obtain spots with maximal, approximately medium and minimal temperatures. Then by using these temperatures we analytically obtain the temperatures all of spots of the specimen and localize the regions consisting of spots of relatively high temperatures. We consider such a region as one to be cracked firstly. This approach allows us to handle only those spots that are of interest and to work in real-time even by using an infrared (IR) camera and a computer with average technical features. We are using the result obtained in this study for fatigue testing the steel materials and for sensing the pre-fatigue state of a specific part of a machine being worked in order to take preventive measures before it breaks down.

Thermographical Investigation of Crack Initiation Using Artificial Neural Networks

In this study, a thermographic infrared imaging system was used to detect the temperature rise of AISI37 steel specimen under reverse bending fatigue. Fatigue behavior of metals shows temperature profiles with three stages: an initial increase of the specimen mean temperature region, a constant (equilibrium) temperature region, an abrupt temperature increase region at end of which the specimen fails and its temperature falls instantly. In order to recognize critical third region, it is necessary to observe endurance state of the specimen being tested. In this study, the temperature profiles of the specimen under testing are recorded by thermal camera and transferred to the image processing program. The artificial neural networks obtain spot temperatures of the inspected specimen by using its temperature profiles. By analyzing the values of obtained data, we detect spots of highest temperatures as ones that are exposed to most intensive deformation. These regions considered to be probable crack initiation sites.

Temperature increase during orthodontic bonding with different curing units using an infrared camera

Abstract Aim. To evaluate the effects of different curing units and light-tip tooth surface distances on the temperature increase generated during orthodontic bonding, using an infrared camera (IR) and artificial neural networks (ANN). Materials and methods. Fifty-two freshly extracted human premolar teeth were used. Metallic orthodontic brackets were bonded to the buccal surfaces of the teeth and thermal records were taken using an IR camera and ANN. Brackets were cured with a light-emitting diode (LED) and high intensity halogen (HQTH). Teeth were divided into four groups according to the curing units (LED and HQTH) and curing distances (from tooth surface and 10 mm away from tooth surface). The results were analyzed with analysis of variance (ANOVA) and the Tukey HSD test. Results. The ANOVA and Tukey HSD tests revealed that temperature changes were influenced by the type of light source and exposure times. All groups revealed significant differences between each other (p < 0.001). The highest surface temperature increase was gained from curing with a LED unit from the tooth surface (11.35°C ± 0.91°C). The lowest surface temperature increase was gained from curing with a HQTH unit 10 mm away from the tooth surface (2.57°C ± 0.6°C). Conclusion. The LED unit induced significantly higher temperature changes than did the HQTH. The temperature increase during orthodontic bonding was increased with long exposure time. A shorter light-tip tooth surface distance leads to greater increases in temperature.

The logic transformations for reducing the complexity of the discernibility function-based attribute reduction problem

The basic solution for locating an optimal reduct is to generate all possible reducts and select the one that best meets the given criterion. Since this problem is NP-hard, most attribute reduction algorithms use heuristics to find a single reduct with the risk to overlook for the best ones. There is a discernibility function (DF)-based approach that generates all reducts but may fail due to memory overflows even for datasets with dimensionality much below the medium. In this study, we show that the main shortcoming of this approach is its excessively high space complexity. To overcome this, we first represent a DF of

Determination of the Fatigue State of the Material under Dynamic Loading by Thermal Video Image Processing

n

The Designing of an Educational Solar Panel That Can Be Controlled in Different Ways

n attributes by a bit-matrix (BM). Second, we partition the BM into no more than

Effect of Hardness upon Temperature Rise of Steel Specimens during Bending Fatigue by Using Infrared Thermography

n-1