Study on Online Recognition Method of Renewable Energy Cascading Tripping Evaluation Based on Machine Learning

Abstract

Considering that there are many factors affecting renewable energy cascading tripping, it is difficult to assess renewable energy tripping risk quickly and accurately. Machine learning has been extensively studied in power system security and stability evaluation. Machine learning can ensure both accuracy and rapidity of renewable energy cascading tripping identification. A method of cascading tripping evaluation based on support vector machine considering conservative principle is proposed. The method combines causal analysis and statistical theory to extract key characteristic quantities. The mapping relationship between system characteristic quantities and renewable energy tripping is established by training to identify cascading renewable energy tripping under contingency and update the prediction model rolling with simulation results to avoid the occurrence of misjudgment. The validity and practicality of the proposed method is verified by an example of actual power system.