A Two-Tier Control Algorithm Based on XGBoost Model to Speed up Smart-Grids Management

Since most of holistic centralized and decentralized methods do not address the problems related to speed in smart-grids, deep-learning and machine-learning methods can be employed outstanding ways to reinforce the stability of power systems with high speed and accuracy. This paper introduces a two-tier control algorithm based on a regression supervised-learning idea to meet both above-mentioned requirements in a group of smart-grids and inverter-based distributed power systems. In this, the power and load managements alongside the supervised learning idea are employed to rapidly detect correlations between real-power, voltage, and frequency. To prove the performance of the proposed algorithm, the modified standard IEEE 37-bus smart-grid is considered, where its steady-state and dynamic results indicate noteworthy effects of that.