Ümmühan Başaran Filik

A smart grid (SG) is an advanced electrical power network that integrates digital technologies and modern communication infrastructures to monitor, control, and manage the flow of information and electricity across generation, transmission, distribution, and consumption systems. The increasing penetration of renewable energy resources, together with the growing demand for rapid and adaptive demand-response mechanisms, has significantly increased the complexity of SG operation and control. Consequently, conventional autonomous control strategies often require accurate system models and substantial computational resources, making them difficult to implement and optimize in large-scale power systems with multiple control regions and dynamic operating conditions. To address these challenges, this study proposes a data-driven multi-agent control framework based on deep reinforcement learning (DRL) for SG applications. As a representative use case, cyberattack detection is investigated by formulating the problem as a multi-agent Markov game within a reinforcement learning framework. To solve the resulting optimization problem, a two-stage multi-agent learning approach is developed, in which the Markov game formulation is enhanced through the integration of differential equation modeling and differential evolution (DE) optimization techniques. The proposed framework enables agents to learn optimal operational policies directly from interactions with the environment, allowing them to effectively distinguish between normal operating conditions and maliciously manipulated data. The effectiveness of the proposed approach is validated through extensive case studies conducted on a SG test system under various operating scenarios, including load and generation fluctuations, uncertainty factors, and cybersecurity vulnerabilities within the communication infrastructure. The results demonstrate that the proposed framework can enhance cyberattack detection performance, improve system resilience, and support the secure and reliable operation of modern SGs.

Natural disasters can cause severe damage to power systems, leading to outages with economic and societal impacts. Ensuring supply of critical loads is therefore a requirement for mitigating the consequences of such events. Microgrids (MGs) can operate in islanded mode, ensuring electricity supply to critical loads through localized energy resources. This study focuses on the formation of MGs by optimally placing distributed generation (DG) units and energy storage systems (ESSs) during outages. The problem is formulated as a mixed-integer conic programming (MICP) model, explicitly accounting for seasonal variations in load demand and season-dependent generation profiles of different DG technologies. The objective function minimizes the total cost, while prioritizing critical loads based on their importance. The model determines the optimal locations and capacities of DGs and ESSs while satisfying the power balance, operational, and topological constraints required for MG formation. The optimization is performed over a 24-hour horizon to capture temporal variations. The framework is applied it to the IEEE 33-bus test system under different seasonal conditions representing the occurrence of a disaster in various periods of the year. Simulation results demonstrate that incorporating seasonal demand and renewable generation characteristics significantly influences DG and ESS allocations.

Ensuring cost-effective and comfortable operation of residential heating, ventilation, and air conditioning (HVAC) systems is crucial for both end-users' financial well-being and thermal comfort. Furthermore, the growing popularity and affordability of on-site energy generation and storage facilities in residential premises offer a potential solution to alleviate the demand for electricity from the grid. Hence, this study is established upon two primary objectives: (i) to establish a desirable equilibrium between the cost of HVAC usage and the avoidance of thermal discomfort among end-users within a multi-objective optimization framework, incorporating a constraint decoupling strategy enabling the real-time HVAC load scheduling, and (ii) to efficiently allocate power generated by on-site sources and stored energy in a home microgrid, utilizing a real-time bi-directional demand response scheme that minimizes costs without compromising the thermal comfort of end-users. This study's validation encompasses various end-user profiles and scenarios, with the results demonstrating the model's capacity to accomplish a favorable balance between cost and comfort for end-users. Furthermore, the implementation of a home microgrid leads to a substantial reduction in daily electricity expenses for end-users from consumer perspective, while it also contributes to a notable decrease in the grid's peak-to-average electricity demand ratio from the utility perspective.

In this study, some of the commonly used daily global solar radiation decomposition models are improved by considering ambient temperature, relative humidity, and air pressure to estimate hourly global solar irradiance from measured daily global solar radiation in Eskişehir, Turkey. Accordingly, the identified functions of the considered parameters are integrated into the selected daily global solar radiation decomposition models. The accuracy of the improved models is evaluated based on the measured monthly average hourly global solar irradiance values. In addition, the monthly average power values provided by the original and improved models are compared with the measured power output values. The results indicate that the improved models have significantly higher performance in the accuracy of both global solar radiation estimation and power output forecasting in a specific time interval. Therefore, the proposed methodology can be used to improve any global solar radiation decomposition model based on locally measured data for different regions.