摘要
为改善现有混合配电网安全分析时数据纬度高、特征复杂导致的安全分析效率低、精度低等问题,提出了一种基于计算智能的混合电网数据分析方法。首先,提出基于快速相关滤波器(FCBF)进行特征选择的方法,从而降低数据维度、提高训练效率。其次,提出一种基于堆叠稀疏去噪自动编码器(SSDAE)的特征提取方法,从而提高算法执行效率及模型鲁棒性。最后,考虑到混合电力系统中存在数据缺失、数据不均衡等问题,提出基于生成对抗网络(GAN)的数据增强方案,从而增强数据多样性、提高模型泛化能力。在试验阶段,以中国某市互联输电网线路采集的数据为例,对所提方法进行了分析和验证。试验结果表明,所提方法训练集和测试集平均相对误差和均方根误差分别为0.0189和0.0209。仿真结果表明,所提方法可以利用未标记样本有效提高回归精度,为电力数据安全分析提供了借鉴。
To improve the problems of low efficiency and low accuracy of security analysis caused by high data latitude and complex features in the existing security analysis of hybrid distribution grids,a computational intelligence-based data analysis method for hybrid grids is proposed.Firstly,a feature selection method based on fast correlation-based filter(FCBF)is proposed to reduce the data dimension and improve the training efficiency.Secondly,a feature extraction method based on stacked sparse denoising automatic encoder(SSDAE)is proposed to improve the algorithm execution efficiency and model robustness.Finally,considering the problems of missing data and data imbalance in hybrid power systems,a data enhancement scheme based on generative adversarial network(GAN)is proposed to enhance data diversity and improve model generalization capability.In the experimental phase,the proposed method is analyzed and validated using data collected from interconnected transmission grid lines in a city in China as an example.The test results show that the average relative error and root mean square error of the proposed method in the training sets and test sets are 0.0189 and 0.0209,respectively.The simulation results show that the proposed method can effectively improve the regression accuracy by utilizing unlabeled samples,which provides reference for the analysis of power data security.
作者
纪素娜
吴丹妍
林幕群
黄朝凯
JI Suna;WU Danyan;LIN Muqun;HUANG Chaokai(Shantou Power Supply Bureau,Guangdong Power Gird Co.,Ltd.,Shantou 515041,China)
出处
《自动化仪表》
CAS
2023年第11期69-73,共5页
Process Automation Instrumentation
基金
南方电网公司科技基金资助项目(030500KK52200003)。
关键词
电力系统
电力调度
遗传算法
目标函数
特征选择
特征提取
数据增强
Power system
Power dispatching
Genetic algorithm
Objective function
Feature selection
Feature extraction
Data enhancement
作者简介
纪素娜(1980—),女,学士,高级经济师,主要研究方向为电力营销信息化、电能量数据应用、营销稽查及营销质量评价管理,E-mail:bhgryyk3377@163.com。
