Modern Methods for Assessing the Reliability of Power Systems under Renewable Energy Integration

Abstract

Relevance: The large-scale utilization of renewable energy sources, particularly wind and solar power plants, has increased the importance of assessing reliability in power systems. Traditional deterministic approaches are insufficient under such conditions, thereby necessitating the application of modern methods based on probabilistic modeling, simulation, and artificial intelligence. These methods enable more accurate evaluation of system stability and efficiency under high variability and uncertainty.

Objective: The main purpose of this study is to analyze modern methods for assessing the reliability of power systems with a high share of renewable energy sources, to identify their advantages and limitations, and to demonstrate the possibilities of applying integrated approaches in practice.

Methods: The study employed probabilistic modeling (Monte Carlo simulation, Markov chains), optimization and risk analysis, time-series-based dynamic simulations, as well as artificial intelligence algorithms (neural networks, ensemble models, anomaly detection methods). System reliability was assessed using indices such as LOLP, LOLE, and EENS. Additionally, the system’s post-disturbance recovery capability was measured by the resilience function:  where represents system performance at time , and is the pre-disturbance performance level.

Results: The results showed that probabilistic approaches provide more accurate reliability indicators compared to classical deterministic models. AI-based forecasting methods helped reduce LOLP values by up to 20–25%, while resilience-oriented approaches shortened system recovery times after disturbances by 15–20%. Moreover, hybrid approaches (probabilistic + AI + resilience metrics) demonstrated the highest effectiveness and proved suitable for integration with real-time monitoring systems.