Hakkari Üniversitesi Akademik Veri Yönetim Sistemi
Not a member yet
6688 research outputs found
Sort by
Forecasting Natural Gas Consumption in Hakkari Province Using Artificial Neural Network
No full textNatural gas consumption is a continuous process that requires an exact supply to prevent interruptions in domestic use, particularly under harsh climatic conditions. This study is presentedto forecast natural gas consumption in Hakkari Province using the Multiple Linear Regression (MLR) and Artificial Neural Network (ANN)modelsfor the first 9 months of 2023. Training data consisted of relative humidity, sunshine duration, temperature, solar intensity, radiation, and subscriber numbers for the 2020-2022 period. The consumption was selected as the dependentoutputparameter, whiletheother parameters were independentinputvariables. Data analysis and training were performed using Python in the Visual Studio Code (VSCode) environment.Data processing was conductedusing the Min-Max normalization method to ensure data consistency, and the 10-fold cross-validation technique was applied.A fully connected feed-forward neural network was usedin the ANN model, whichconsisted of three hidden layers containing 32, 16, and 8 neurons.LeakyReLU was preferred as the activation function in the hidden layers to providenon-linearity, while a linear activation function was implementedin the output layer for regression.The Adaptive Moment Estimation (Adam) algorithm was selected as the optimizer to update the network weights and minimize the loss function more efficiently. The R square,RMSE and MAE values were found as 0.755,0.128, 0.104for the MLR model, and 0.943,0.06, 0.034for the ANN model, respectively.The results indicatethat the ANN model showed betterperformancein terms of forecasting compared to the MLR model
Great Power Cyberpolitics and Global Cyberhegemony
No full textAs interstate cyberconflict intensifies, the intersection of national security, cybersecurity, and International Relations (IR) theory has emerged as a critical venue for scholarly inquiry. Yet due mainly to epistemological problems, IR theory has been limited in examining how it informs the maximization of strategic cyberpower and in testing key realist concepts and assumptions against cyberactualities, risking theoretical stagnation and conceptual infertility in the study of statecraft and cybersecurity. I seek to bridge these theory-testing and conceptual gaps by assessing offensive realism's assumption about the scope of hegemonic expansion in cyberspace using the crucial case of the United States. I argue that offensive realism has meaningful explanatory and predictive power in cyberspace but sometimes lacks this power under conditions assumed by the theory, emphasizing the need to modify offensive realism's understanding and scope conditions of hegemony. The US pursues global, not regional, cyberhegemony using offensive strategies to maximize its cyberpower for cybersecurity. Therefore, I critically examine defensive realism and cyber persistence theory as alternative structural perspectives on the pursuit of security in cyberspace and introduce a modified conceptual framework for hegemony to adapt offensive realism to cyber-realities. This conceptual innovation can potentially contribute to policy making and help to build a cyber-specific version of offensive realism
Comparative Analysis of PID and Artificial Neural Network-Based Controllers for Attitude Determination and Control in Spacecraft
No full textSpacecraft's attitude determinationand control systems (ADCS) need high-performance controllers since they need tobe able to move very precisely and because the actuators don't have a lot ofpower. This research examines the attitude control problem of a rigidspacecraft, comparing the numerical performance metrics of a traditionalProportional-Integral-Derivative (PID) controller with those of an ArtificialNeural Network (ANN)-based intelligent controller. We used Euler angles andangular velocities to make the spacecraft's dynamic model, and we used a rigidbody dynamics approach to simulate moments of inertia. The PID controller isset up to work with reference moves, while the ANN-based controller is taughtto learn how the system behaves in a nonlinear way.The simulation findings showthat when using a PID controller, the settling time for the orientation erroris about 8–10 seconds. However, when utilizing an ANN-based controller, thistime is cut down to 4–5 seconds. The greatest overshoot with a PID controlleris about 12–15%, however with an ANN-based controller, this number has droppedto less than 5%. Also, adopting an ANN-based controller cut steady-stateinaccuracy by about 40–50%. When looking at the control torque needs, it wasfound that the PID controller needed greater instantaneous torque, while theANN-based controller made a smoother and more energy-efficient control signal.The results clearly reveal that the PID controller is better since it is easierto use and works in more situations. The ANN-based controller, on the otherhand, works better when there are nonlinearities and parameter uncertainties.</p
