A Discrimination Method of Power System Dominant Instability Mode Based on Active Learning

Abstract

The invention discloses a method for discriminating the dominant instability mode of a power system based on active learning, which belongs to the field of power system stability analysis. The present invention applies active learning to the method for discriminating the dominant instability mode, without labeling all samples, but only actively selects some samples with the most information content for labeling, and achieves higher discrimination accuracy with fewer samples for labeling. At the same time, on the basis of model uncertainty, the diversity of samples is also considered, and the model output probability information entropy is used as the weight, and the samples are clustered into k clusters by weighted k-means clustering method, and the closest ones are selected from each cluster. Clustering the samples of centroids ensures the diversity of sampling and avoids the problem of over-fitting of the model caused by the selected samples being too similar. In this way, the present invention can reduce the cost of sample labeling, improve the realization efficiency of the dominant instability mode discrimination model, thereby improving the safety of the power system, and has strong practicability.

Description

technical field [0001] The invention belongs to the field of power system stability analysis, and more specifically relates to a method for discriminating the dominant instability mode of a power system based on active learning. Background technique [0002] The safe and stable operation of the power system is crucial to national energy security and economic and social development, and its security and stability are the key concerns of system planning, operation, and protection workers. Quickly and accurately judging the stability and instability of the power system and the dominant instability mode (voltage instability, power angle instability) after the power system is subject to a large disturbance will buy time for emergency control measures and provide a basis for which measures to take , so as to effectively ensure the security and stability of the system. [0003] Time-domain simulation, energy function, bifurcation analysis, etc. are commonly used methods to identif...

AI Technical Summary

Problems solved by technology

[0005] However, existing research uses supervised learning methods when training deep learning models, which requires a large number of labeled samples

In fact, the labeling of the dominant instability mode identification samples cannot be obtained directly through the simulation program, but needs to be judged and labeled manually based on the experience of power grid experts, which takes a lot of time and labor costs, and affects the efficiency and applicability of the model.

Images