基于改进随机森林的电力系统暂态稳定评估

针对传统基于机器学习的电力系统暂态稳定评估方法存在准确率偏低和泛化能力不足的问题,提出了一种基于特征选择和改进随机森林的在线暂态稳定评估方法。首先,通过最大化联合互信息挖掘电网运行数据之间的相关性,筛选出具有代表性的关键特征子集;然后,考虑到电力系统数据库中稳定样本与失稳样本之间的类别不平衡问题,通过改进bootstrap抽样和对决策树进行加权处理,增强随机森林对失稳样本的识别能力;最后,基于改进的随机森林算法,建立关键特征数据与暂态稳定标签之间的映射关系。实验结果表明,所提方法具有较高的准确性和较强的鲁棒性,能够满足在线应用的需求。

Transient Stability Assessment of Power System Based on Improved Random Forest

To solve the problems of low accuracy and insufficient generalization ability of traditional power system transient stability assessment methods based on machine learning, an online transient stability assessment method based on feature selection and improved random forest was proposed. Firstly, the joint mutual information maximisation was used to mine the correlation between grid operation data to select the representative key feature subset. Then, considering the class imbalance between stable samples and unstable samples in the database of the power system, the ability of the random forest to identify unstable samples was enhanced by improving bootstrap sampling and weighting the decision tree. Finally, the mapping relationship between key feature data and transient stability label was established based on the improved random forest algorithm. The test results show that the proposed method has high accuracy and strong robustness, which can meet the requirements of online application.