基于迁移学习的OMS配网瞬态稳定性控制方法  

Transient stability control method of OMS distribution network based on Transfer Learning

作　　者：邝梓佳 邱桂华 汤志锐 KUANG Zi-jia;QIU Gui-hua;TANG Zhi-rui(Guangdong Power Grid Company Foshan Power Supply Bureau,Foshan 528000,Guangdong Province,China)

机构地区：[1]广东电网有限责任公司佛山供电局,广东佛山528000

出　　处：《信息技术》2024年第1期182-188,共7页Information Technology

摘　　要：为了提升一体化控制效果,确保OMS配网的瞬态稳定,研究基于迁移学习的OMS配网瞬态稳定性控制方法。通过迁移学习调整网络中非固定层参数,依据提取的特征训练重构非固定层的网络,构建迁移学习卷积神经网络模型;输入二维电压电流数据图像,输出OMS配网瞬态稳定性预测结果;若OMS配网瞬态失稳,则求解OMS配网内发电机的灵敏度,按照灵敏度生成OMS配网瞬态稳定性一体化控制策略。实验证明,该方法在不同故障下,综合评价指标值较高,一体化控制效果较佳。In order to improve the integrated control effect and ensure the transient stability of OMS distribution network,the transient stability control method of OMS distribution network based on transfer learning is studied.The parameters of the non-fixed layer in the network are adjusted by transfer learning,the network of the non-fixed layer is reconstructed according to the extracted feature training,and the transfer learning convolution neural network model is constructed.The two-dimensional voltage and current data image is input and the transient stability prediction results of OMS distribution network is output.In case of transient instability of OMS distribution network,the sensitivity of generators in OMS distribution network is solved,and the integrated control strategy of transient stability of OMS distribution network is generated according to the sensitivity.Experiments show that under different faults,the comprehensive evaluation index value is higher and the integrated control effect is better.

关 键 词：迁移学习 OMS配网 瞬态稳定性 一体化控制 灵敏度 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]