Methodology for Forming a Training Sample for Power Systems Emergency Control Algorithm Based on Machine Learning

The current stage of power systems development is characterized by changes in the fundamental principles of electrical modes planning and control. These changes are associated with the active introduction of renewable energy sources, flexible AC transmission systems based on power electronics and changes in the rules for the operation of electricity markets. As a result, modern power systems are characterized by a significant uncertainty in electrical modes and a high rate of transients processes as well. In such conditions of power systems operating, a number of problems arise in the operation of traditional emergency control devices. A number of systemic accidents have shown that the new operating conditions of modern power systems lead to the need to increase the speed and adaptability of emergency control devices. Today, this problem can be solved by using machine learning algorithms. The article presents a technique for forming a training sample for power systems emergency control algorithm based on machine learning. Testing was performed for multivariate classification algorithms.