Data-driven stability assessment of power systems

Abstract

In our nation's economic progression, the power system plays a pivotal role. As we advance and the scale of our power systems grows, their operational complexity multiplies. Random disturbances, faults, and the instability inherent in power electronic devices pose significant challenges to the safe and stable functioning of AC/DC hybrid grids. To address the escalating demand for electricity, modern power systems are evolving towards higher voltages, increased capacity, and extended transmission distances. Furthermore, the integration of a significant portion of renewable energy has made the grid structure significantly more complex. Analyzing power system stability, utilizing data-driven approaches to gather and analyze stability data, is crucial for enhancing voltage stability and preventing widespread power outages. This research is vital for ensuring the resilience and reliability of our electrical infrastructure. This paper is based on the constructed voltage stability margin of the power system, employing artificial intelligence neural networks and mathematical models for optimization problems to conduct an in-depth study on emergency control strategies for grid voltage. It aims to explore emergency control strategies for grid voltage stability based on artificial intelligence methods to enhance the operational safety and reliability of power systems. In the section on artificial intelligence neural networks, for two different types of data under various conditions, three models based on artificial intelligence neural networks for assessing grid voltage stability were compared. Through data analysis, these models enable rapid and accurate predictions of grid voltage stability, providing crucial references for the formulation of subsequent control strategies. The results show that artificial intelligence neural networks perform well in predicting grid voltage stability and can employ different artificial intelligence models to predict different types of grid data. The study concludes that among the two types of power system data studied, different models exhibit different performances for various data sets. The use of artificial intelligence in researching emergency control strategies for grid voltage has proven to have good accuracy and feasibility. However, achieving higher accuracy and faster response rates simultaneously may not be feasible, and in large-scale grid control, this could potentially cause more severe impacts. This research provides simulation results for different emergency situations in the grid through artificial intelligence methods, offering high accuracy and significant engineering relevance.