Denis A. Boyarkin, Dmitriy S. Krupenev, Dmitrii V. Iakubovskii

Melentiev Energy Systems Institute Siberian Branch of the Russian Academy of Sciences

The article considers the question of increasing the computational efficiency of the methodology for electric power system adequacy assessment based on the Monte Carlo method. When using this method, the speed and accuracy of the calculation depends on the count of analyzed random states of the modeled system. Analysis means the solution of the flow distribution problem for each randomly generated state. This is a fairly time-consuming process, so reducing the number of analyzed states while maintaining the accuracy of the assessment will increase the speed of the calculation as a whole. For this, it is proposed to use methods of machine learning, whose task is to determine the power system's deficit without optimization methods applying. In calculations, the support vector machine and random forest methods were used, the viability of the proposed approach was evaluated by solving adequacy assessment task of the test energy power system

electric power systems, adequacy assessment, Monte Carlo method, machine learning