Scholarlink

<p>The increasing adoption of renewable energy resources (RERs) in microgrids (MGs) poses new challenges for power system frequency stability, as inverter-dominated grids lack the inherent inertia of traditional synchronous generation. To conduct frequency stability studies on power systems, it is essential to have a dynamic equivalent (DEM) model of the connected MG that can accurately reflect its true dynamic behavior during power system disturbances. In this paper, to understand the MG’s dynamic behavior with high penetration of RERs during transient frequency, the dynamic modeling of the MG is developed. This model aims to reduce computational complexity and alleviate the workload in research related to dynamic frequency stability using deep neural network (DNN) approach. In the proposed method, different levels of RER penetration in microgrid, different power system loading level, and frequency variations are considered as inputs and active power changes injected from the MG to the power system is considered as the output of the DNN model. The GridSearchCV approach has been employed to fine-tune the hyperparameters, which include the number of neural network layers, the number of neurons, and the activation functions for each layer. The results obtained using the DigSILENT Power Factory demonstrate that the DEM achieves &gt; 99% prediction accuracy (R² = 0.9979) for active power changes across all scenarios, with a mean absolute error (MAE) of 0.0359 pu and root-mean-square error (RMSE) of 0.0674 pu. These findings imply that the performance of the equivalent model aligns consistently with the actual model.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Dynamic Equivalent Modeling of Microgrids with High Penetration of Renewable Energy Resources for Power System Frequency Stability Studies Using Deep Neural Networks (DNNs)

  • 期刊论文

  • 发表日期：

    2025年9月1日

查看全文
  • Masoumeh Rezazadeh Seylab,
  • Mehdi S. Naderi,
  • Gevork B. Gharehpetian

摘要

The increasing adoption of renewable energy resources (RERs) in microgrids (MGs) poses new challenges for power system frequency stability, as inverter-dominated grids lack the inherent inertia of traditional synchronous generation. To conduct frequency stability studies on power systems, it is essential to have a dynamic equivalent (DEM) model of the connected MG that can accurately reflect its true dynamic behavior during power system disturbances. In this paper, to understand the MG’s dynamic behavior with high penetration of RERs during transient frequency, the dynamic modeling of the MG is developed. This model aims to reduce computational complexity and alleviate the workload in research related to dynamic frequency stability using deep neural network (DNN) approach. In the proposed method, different levels of RER penetration in microgrid, different power system loading level, and frequency variations are considered as inputs and active power changes injected from the MG to the power system is considered as the output of the DNN model. The GridSearchCV approach has been employed to fine-tune the hyperparameters, which include the number of neural network layers, the number of neurons, and the activation functions for each layer. The results obtained using the DigSILENT Power Factory demonstrate that the DEM achieves > 99% prediction accuracy (R² = 0.9979) for active power changes across all scenarios, with a mean absolute error (MAE) of 0.0359 pu and root-mean-square error (RMSE) of 0.0674 pu. These findings imply that the performance of the equivalent model aligns consistently with the actual model.