| 引用本文: | 卫志农,李超凡,丁爱飞,孙国强,黄蔓云,臧海祥,方熙程.基于Tri-training-SSAE半监督学习算法的电力系统暂态稳定评估[J].电力自动化设备,2023,43(7): |
| WEI Zhinong,LI Chaofan,DING Aifei,SUN Guoqiang,HUANG Manyun,ZANG Haixiang,FANG Xicheng.Power system transient stability assessment based on Tri-training-SSAE semi supervised learning algorithm[J].Electric Power Automation Equipment,2023,43(7): |
|
| 摘要: |
| 基于机器学习的暂态稳定评估方法主要采用监督学习方法，为了解决监督学习方法所需的有标签样本难以获取的问题，提出基于三体训练-稀疏堆叠自动编码器(Tri-training-SSAE)半监督学习算法的电力系统暂态稳定评估方法。构建基于堆叠稀疏自动编码器的暂态稳定评估模型；在传统的三体训练过程中加入伪标签样本置信度判断，以减小噪声数据对模型训练的影响；以堆叠稀疏自动编码器为基分类器构建三体训练-稀疏堆叠自动编码器模型，利用大量的无标签样本提高模型的泛化能力。通过IEEE 39节点系统与华东某省级电网进行分析验证，结果表明，所提方法在有标签样本数较少时具有更高的评估准确度。 |
| 关键词: 暂态稳定评估 机器学习 半监督学习 三体训练算法 堆叠稀疏自动编码器 |
| DOI：10.16081/j.epae.202212009 |
| 分类号:TM712 |
| 基金项目:国家自然科学基金资助项目(U1966205) |
|
| Power system transient stability assessment based on Tri-training-SSAE semi supervised learning algorithm |
|
WEI Zhinong1, LI Chaofan1, DING Aifei2,3, SUN Guoqiang1, HUANG Manyun1, ZANG Haixiang1, FANG Xicheng4
|
|
1.College of Energy and Electrical Engineering, Hohai University, Nanjing 211100, China;2.NARI Jidian New Energy(Nanjing) Co.,Ltd.,Nanjing 211106, China;3.NARI Technology Co.,Ltd.,Nanjing 211106, China;4.Yangzhong Power Supply Company of State Grid Jiangsu Electric Power Co.,Ltd.,Zhenjiang 212200, China
|
| Abstract: |
| The transient stability assessment methods based on machine learning mainly use supervised learning method, in order to solve the problem that it is difficult to obtain the labeled samples needed by supervised learning method, a power system transient stability assessment method based on Tri-training-stacked sparse auto-encoder(Tri-training-SSAE) semi supervised learning algorithm is proposed. A transient stability assessment model based on stacked sparse auto-encoder(SSAE) is constructed. The pseudo label sample confidence judgment is added to the traditional Tri-training process to reduce the impact of noise data on model training. The SSAE is taken as the base classifier, a Tri-training-SSAE model is constructed, and a large number of unlabeled samples are used to improve the generalization ability of the model. The analysis and verification are carried out through IEEE 39-bus system and a provincial power grid in East China, and the results show that the proposed method has higher assessment accuracy rate when the number of labeled samples is small. |
| Key words: transient stability assessment machine learning semi supervised learning Tri-training algorithm stacked sparse auto-encoder |
