A Model-Free Approach for Assessing Renewable Generation Penetration in Power Systems

Abstract

Abstract—Integrating variable renewable energy (VRE) gen- eration into power systems presents significant challenges due to its inherent variability and uncertainty. Traditionally, the impact of this generation is assessed using power flow analysis tools. However, this method requires modeling the characteristics of the electrical grid, and its solution can entail excessive computational effort for real-time applications. In this regard, methods based on artificial intelligence have emerged as an alternative for operating electrical power systems. Therefore, this paper proposes a strategy based on a multilayer perceptron (MLP) to assess the impact of VRE generation, such as wind and solar, through a purely data-driven approach. To evaluate the proposed strategy, numerous operating scenarios were simulated on the IEEE 30-bus test system using a power flow tool, and the resulting solutions were used to train the MLP. Once the MLP was trained, the system voltage profiles could be estimated with a mean squared error (MSE) of approximately 1.47 ×10^-5 and a coefficient of determination (R-squared) of 0.987 on the test set. These findings provide a scalable and adaptable tool for system operators to effectively manage the increasing penetration of VRE generation.