SVG Loss Estimation Method Based on Extreme Gradient Boosting

Abstract

Abstract Static Var Generator (SVG) consumes large amounts of electrical energy during its operation, generating high equipment operating costs. Therefore, it is necessary to quantify the operating SVG loss. However, the existing methods of loss estimation generally have notable problems, such as large deviations in loss estimation results. In this paper, we propose a method for calculating SVG loss and establish an SVG loss estimation model based on the extreme gradient boosting algorithm under different operating conditions. First, some data obtained by simulation experiments are input into the algorithm model for training, and the hyperparameters of the model are adjusted by the Bayesian optimization algorithm. Then, a new test set is constructed by superimposing the measurement errors with different signal-to-noise ratio noises on the test set data. Finally, the test set is fed into the trained model for testing to validate the effect of the present model. The experimental results show that the model can quickly and accurately achieve loss estimation of SVG devices with good engineering practicality.