Modeling and Design Optimization to Determine the Mechanical Properties of a Recent Composite

Naciye Burcu KARTAL

Makale | 2021 | Journal of Artificial Intelligence and Data Science1 ( 1 ) , pp.80 - 88

This study proposes an appropriate optimization model for determining a new composite material's mechanical properties by neuro-regression analysis. This new composite material is obtained by combining hemp and polypropylene fibers. It was developed for the sector of upholstered furniture. First, different multiple regression models have been tried for input and output values. The R2 training, R2 testing, R2 validation, and minimum, maximum values were determined for each model. Then, the stochastic optimization approach is used to predict and optimize the mechanical properties of the new biocomposite system. Finally, multiple non-l . . .inear models determine the maximum tensile strength and elongation achievable within the constraints. It is found what the optimum input parameters are needed to achieve maximum tensile strength and elongation at break values of the material and that the type of scenario and the choice of constraints for design variables are critical in the optimization problem Daha fazlası Daha az

Modeling and optimum design for wire electrical discharge machining of γ titanium aluminide alloy


Makale | 2021 | Journal of Artificial Intelligence and Data Science1 ( 1 ) , pp.89 - 95

Wire electrical discharge machining (WEDM) of γ titanium aluminide is the subject of the current research. Due to the large number of process variables and sophisticated stochastic process mechanisms, selecting the best machining parameter combinations for increased cutting efficiency and accuracy is a difficult task in WEDM. In general, there is no perfect combination that can produce the fastest cutting speed and the finest surface finish quality at the same time. For this purpose, the data were selected from a literature study. This study describes an attempt to devise a suitable machining technique for achieving the highest poss . . .ible process criteria yield. To model the machining process, a stochastic optimization method, differential evolution, has been performed. Cutting speed, surface roughness, and wire offset are the three most important criteria that have been used as indicators of process performance. The response characteristics can be predicted as a function of six different control parameters, namely pulse on time, pulse off time, peak current, wire tension, dielectric flow rate, and servo reference voltage. The limitations of the candidate models are checked after the R 2 training, R2 testing and R2 valiadtion values are calculated to reveal whether the model is realistic. Optimization results are 3.02 mm/min, 2.36 µm, and 0.13 mm for the maximum cutting speed, the minimum surface roughness, and minimum wire offset, respectively. It is shown that the machining model is suitable and that the optimization technique meets practical requirements Daha fazlası Daha az

Fabric Defect Classification Using Combination of Deep Learning and Machine Learning


Makale | 2021 | Journal of Artificial Intelligence and Data Science1 ( 1 ) , pp.22 - 27

Automatic systems can be used in many areas, such as the production stage in factories, country defense, and traffic control. They provide the opportunity to reach results faster with higher success rates thanks to human-computer vision cooperation. In this study, it is aimed to develop an intelligent system that automatically detects and classifies defects in fabrics. Thanks to the developed system, the cause of the malfunction is eliminated, and the recurrence of the malfunction is prevented. Using deep learning methods in fabric defect classification studies has a disadvantage compared to other methods. Multiple layers in deep le . . .arning cause a time-consuming process. Therefore, a combination of Deep Learning and Support Vector Machines (SVM) has been used in this study. The success of the provided system has been compared with other deep learning algorithms in terms of time and accuracy Daha fazlası Daha az

A Novel, Nelder-Mead Optimization Approach, based on Neuro-regression modeling for the Energy Efficiency Parameters of End Milling Process


Makale | 2021 | Journal of Artificial Intelligence and Data Science1 ( 1 ) , pp.96 - 105

Global crises are increasing day by day due to the rapid depletion of energy supplies around the planet. One of the goals of engineering is to prevent this situation by developing innovative solutions to this rapid energy consumption that has disappeared in the world. A solution could be to reduce the energy consumption of the machines that are used during production. In this study, a new design technique based on the neuro-regression approach and non-linear regression modeling was offered as an alternative to Taguchi design to reduce energy consumption. Thus, a cutting parameter optimization model was created to examine the effects . . . of the constraint conditions on energy consumption. The cutting power, the surface roughness of the part, and tool life were handled as objective functions(constraint conditions). First of all, the multiple non-linear regression modeling was created using design variables in end milling . These design variables were determined as spindle rotational speed, feed rate power, radial cut depth, axial cut depth, and cutting speed. Then, objective functions were brought to the proper minimum optimal levels due to this optimization modeling. As a result of the optimization model built with design variables, accurate modeling was achieved in this work by studying several optimization models utilized to optimize the minimum objective functions, which play a significant role in reducing energy consumption in end milling. After the optimization, the maximum value was found as 110.791. At the end of the study, some options of direct search method to maximize and minimize results were applied Daha fazlası Daha az

A Flower Status Tracker and Self Irrigation System (FloTIS)

Rumeysa KESKİN | Furkan GÜNEY | M. Erdal ÖZBEK

Makale | 2021 | Journal of Artificial Intelligence and Data Science1 ( 1 ) , pp.45 - 50

The Internet of Things (IoT) provides solutions to many daily life problems. Smartphones with user-friendly applications make use of artificial intelligence solutions offered by deep learning techniques. In this work, we provide a sustainable solution to automatically monitor and control the irrigation process for detected flowers by combining deep learning and IoT techniques. The proposed flower status tracker and self-irrigation system (FloTIS) is implemented using a cloud-based server and an Android-based application to control the status of the flower which is being monitored by the local sensor devices. The system detects chang . . .es in the moisture of the soil and provides necessary irrigation for the flower. In order to optimize the water consumption, different classification algorithms are tested. The performance comparisons of similar works for example flower case denoted higher accuracy scores. Then the best generated deep learning model is deployed into the smartphone application that detects the flower type in order to determine the amount of water required for the daily irrigation for each type of flower. In this way, the system monitors water content in the soil and performs smart utilization of water while acknowledging the user Daha fazlası Daha az

A Face Authentication System Using Landmark Detection

Velican ERCAN | M. Erdal ÖZBEK

Makale | 2021 | Journal of Artificial Intelligence and Data Science1 ( 1 ) , pp.28 - 34

Biometric data is the key for many security applications. Authentication relies on the individual’s measurable biometric properties collected as features. In this study, a face authentication system is built to be used in opening the entrance door accessing to the apartments and housing estates. The proposed system consists of three stages. In the first stage, landmarks on the face are captured using a deep neural network. Then six selected features from the landmarks are extracted and traditional machine learning algorithms are used to authenticate users. In the last stage, a user interface is built. Face recognition tests achieved . . . an accuracy rate of 89.79% Daha fazlası Daha az

6698 sayılı Kişisel Verilerin Korunması Kanunu kapsamında yükümlülüklerimiz ve çerez politikamız hakkında bilgi sahibi olmak için alttaki bağlantıyı kullanabilirsiniz.

Bu site altında yer alan tüm kaynaklar Creative Commons Alıntı-GayriTicari-Türetilemez 4.0 Uluslararası Lisansı ile lisanslanmıştır.