Yazar "Aksoy, Necati" seçeneğine göre listele

Listeleniyor 1 - 6 / 6
  • Küçük Resim Yok
    Öğe
    Deep learning approaches for robust prediction of large-scale renewable energy generation: A comprehensive comparative study from a national context
    (Sage Publications Inc, 2025) Aksoy, Necati; Genc, Istemihan
    Precise forecasting of renewable energy generation is crucial for ensuring grid stability and enhancing the efficiency of energy management systems. This research develops and rigorously evaluates a range of deep learning models-such as Recurrent Neural Networks (RNNs), Long Short-Term Memory (LSTM) networks, Gated Recurrent Units (GRUs), and Bidirectional LSTM (BiLSTM) architectures-for predicting solar, wind, and total renewable energy production at a national scale. These models are systematically benchmarked against traditional machine learning approaches and gradient boosting methods to determine their predictive capabilities. The findings demonstrate that deep learning models incorporating memory mechanisms consistently surpass conventional methods, with BiLSTM standing out as the most precise and dependable model. Furthermore, the study investigates fully connected artificial neural networks (ANNs) and ConvLSTM2D models, reinforcing the advantages of memory-based architectures in modeling temporal relationships. By introducing a robust deep learning framework for large-scale renewable energy forecasting, this research represents a considerable leap forward compared to traditional machine learning techniques. The results highlight the transformative potential of deep learning in improving forecasting accuracy, thereby facilitating more effective energy planning and the smooth integration of renewable energy into national power grids.
  • Küçük Resim Yok
    Öğe
    Eliminating Meteorological Dependencies in Solar Power Forecasting: A Deep Learning Solution With NeuralProphet and Real-World Data
    (Ieee-Inst Electrical Electronics Engineers Inc, 2025) Aksoy, Necati; Yilmaz, Alper; Bayrak, Gokay; Koc, Mehmet
    Forecasting solar power generation is essential for efficient energy management and grid stability. However, existing predictive models often rely on external datasets, such as meteorological and sensor data, to make accurate predictions. This dependency introduces complexities and limits their application in data-sparse scenarios. In this study, we propose a novel forecasting approach based on the NeuralProphet algorithm, a deep learning model that predicts solar power generation solely from its historical data, eliminating reliance on additional input data. To evaluate the proposed approach, we conducted two case studies. The first utilized a 10-month dataset from a 1.2 kW small-scale solar power unit at Bursa Technical University's Smart Grids laboratory, recorded at 15-minute intervals. Despite the limited dataset, the model achieved an R-squared value exceeding 0.74, demonstrating promising predictive capability. The second case study applied the NeuralProphet-based model to a large-scale dataset of nationwide solar power generation in Germany, spanning five years and collected at 15-minute intervals. Models trained on this dataset achieved R-squared values exceeding 0.99, highlighting the algorithm's capacity to effectively capture seasonal and temporal patterns at a national scale. Our results indicate that the NeuralProphet-based forecasting approach offers a viable and efficient alternative for solar power prediction, achieving high accuracy without external data dependencies.
  • Küçük Resim Yok
    Öğe
    Optimizing Battery Cooling with Reinforcement Learning: A Dynamic Control Strategy for Energy Storage Systems
    (Ieee, 2025) Aksoy, Necati; Cakil, Fatih; Ozden, Mustafa
    The growing reliance on energy storage systems (ESS) in residential, vehicular, and industrial applications necessitates efficient thermal management solutions to enhance performance and longevity. This study proposes a reinforcement learning (RL)-based cooling control system for optimizing the temperature regulation of battery banks. Unlike traditional rule-based or static cooling strategies, the proposed method dynamically adjusts coolant flow rates using an Expected SARSA agent, which learns an optimal control policy through interactions with a custom-designed environment model. The model accurately simulates thermal absorption and coolant flow dynamics, allowing for precise and adaptive cooling regulation. The performance of the RL agent was evaluated across different training durations, demonstrating that extended training significantly improves stability, reward consistency, and energy efficiency. Compared to conventional cooling strategies, the RL-based system ensures more adaptive, efficient, and reliable thermal management, making it a promising solution for next-generation energy storage applications.
  • Küçük Resim Yok
    Öğe
    Performance Analysis of Deterministic and Probabilistic Path Planning Algorithms in Complex Environments
    (Institute of Electrical and Electronics Engineers Inc., 2024) Aksoy, Necati; Çakil, Fatih; Tekdemir, Ibrahim Gürsu
    Path planning, in an other saying navigation, algorithms are vital in assorted applications, including robotics, autonomous vehicles, and drones. These algorithms can be broadly categorized into deterministic and probabilistic methods along with other branches. This study focuses and examines two classical deterministic path planning algorithms, A-star (A*) and Dijkstra's algorithm, alongside two prominent probabilistic path planning algorithms, Rapidly-exploring Random Trees Star (RRT*) and Probabilistic Roadmap (PRM). In the paper, with creating multi-level building interior floor maps and testing the performance of these four algorithms are performed on each level. Performance metrics included execution time, CPU usage, memory usage, and path distance. The results, presented in comparative tables, provide a comprehensive analysis of the efficiency and resource demands of each algorithm. Furthermore, this research offers valuable insights for selecting appropriate path planning algorithms in various autonomous navigation applications, guiding future implementations in robotics, autonomous electric vehicles, and drone technology. © 2024 IEEE.
  • Küçük Resim Yok
    Öğe
    Real-Time Smart Power Distribution in Electric Vehicle Chargers: Utilizing Rust Programming for Enhanced Efficiency
    (Institute of Electrical and Electronics Engineers Inc., 2024) Çakil, Fatih; Aksoy, Necati; Tekdemir, Ibrahim Gürsu
    The increase in the number of electric vehicles (EVs) and charging units generates a substantial power demand, necessitating the effective management of these loads. This study examines a model that employs the Priority-Based Power Distribution approach to prioritize EVs at charging stations based on real-time power demand. By leveraging the advantages of the Rust programming language, we developed and simulated a real-time model framework. This research introduces an innovative interface and design for controller systems, illustrating the critical role of Rust programming in optimizing EV charging station operations. The results underscore the efficiency and practicality of our proposed model in effectively managing dynamic power distribution. © 2024 IEEE.
  • Küçük Resim Yok
    Öğe
    Reinforcement learning-based multi-objective smart energy management for electric vehicle charging stations with priority scheduling
    (Pergamon-Elsevier Science Ltd, 2025) Cakil, Fatih; Aksoy, Necati
    Reinforcement learning (RL)-based control structures represent a transformative approach to optimizing energy management in electric vehicle (EV) charging stations, offering unparalleled adaptability and efficiency. This paper introduces a novel RL-based intelligent control strategy, designed to address multi-objective challenges in EV charging, such as energy efficiency, cost-effectiveness, and user prioritization. Central to this study is the development of a unique environment model, which incorporates dynamic variables including vehicle priority, arrival times, and real-time pricing data, ensuring realistic and practical applications. Additionally, a custom reward strategy is proposed, enabling the RL agents to effectively learn and adapt to complex operational demands. The study evaluates the performance of three RL algorithms-Q-Learning, SARSA, and Expected SARSA-within the proposed environment model, demonstrating their capabilities in reducing charging costs and improving profitability. Experimental results indicate that the Q-Learning agent achieved an average cost reduction of up to 66.4% compared to conventional charging strategies, with energy costs dropping from 11.78 to 3.96 per unit in high-priority cases. Expected SARSA exhibited a competitive performance, yielding up to 44.3% cost savings, whereas SARSA consistently resulted in the lowest cost efficiency, with reductions of only 46.6% in comparable scenarios. Furthermore, the RL-based scheduling framework successfully shortened peak-hour waiting times by 40%, while ensuring equitable prioritization of charging requests. Through a comprehensive set of realistic case studies and scenarios, the effectiveness of these algorithms is analyzed, focusing on their capacity to manage energy costs, enhance profitability, and adapt to fluctuating pricing conditions.