Power system transient stability discrimination method based on XGBoost algorithm

Abstract

The invention provides a power system transient stability discrimination method based on the XGBoost algorithm. According to the method, firstly, the grid simulation software is utilized to simulate the transient process of a to-be-evaluated grid under various operating modes after node and line faults; electrical quantity characteristics are extracted from the simulation data, and the temporary stability criterion is utilized to determine a temporary stability label; secondly, sample data is then utilized to train an XGBoost model; according to characteristics of different severity degrees oftwo types of errors of transient stability prediction, the attention coefficient is introduced to correct a loss function of the algorithm; the logistic function is utilized to probabilize model output. The method is advantaged in that accuracy and the recall rate are relatively high, difference between relatively-determined prediction and relatively uncertain prediction can be captured in a probabilistic manner, and a part of the false output of the model can be avoided.

Description

Technical field [0001] The invention belongs to the field of power systems, and specifically is a method for judging the transient stability of the power system. Background technique [0002] Due to the improvement of grid interconnection level, increasing load, new energy access, line transmission capacity restrictions and other factors, the operation of the power system is getting closer to its stability limit, and the stable operation of the power grid shows greater importance, thus the problem of transient stability assessment (Transient Stability Assessment, TSA) has attracted more attention. The traditional TSA method based on time domain simulation is limited by the calculation speed, and it is difficult to meet the needs of online applications. In recent years, with the rapid development of artificial intelligence technology, transient stability assessment based on machine learning algorithms has become a research hotspot for scholars. [0003] Machine learning algorithms...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is to overcome the defects of the existing technology, including the modeling accuracy to be further improve

Images