基于LSTM的重要用户电能质量趋势预测分析模型  

Trend Prediction and Analysis Model for Power Quality of Important Users Based on LSTM

作　　者：赵长伟 骈睿珺 杜天硕 葛磊蛟 ZHAO Changwei;PIAN Ruijun;DU Tianshuo;GE Leijiao(Chengdong Power Supply Branch,State Grid Tianjin Electric Power Company,Tianjin 300260,China;School of Electrical and Information Engineering,Tianjin University,Tianjin 300072,China)

机构地区：[1]国网天津市电力公司城东供电分公司,天津300260 [2]天津大学电气自动化与信息工程学院,天津300072

出　　处：《电力系统及其自动化学报》2022年第7期26-33,共8页Proceedings of the CSU-EPSA

基　　金：国网天津市电力公司科技项目(KJ21-1-18)。

摘　　要：为精确掌握重要用户电能质量的变化趋势规律,进而有效实现对其高品质的电能质量供应,文中提出一种基于距离相关系数与长短时记忆LSTM(long short-term memory)网络的重要用户电能质量趋势预测分析方法。首先,对可表征重要电力用户电能质量的多维度特征原始数据进行标准化处理,进一步利用距离相关系数过滤低相关特征实现特征降维,从而完成趋势变化基础样本数据集的筛选;其次,将训练集样本输入到双层LSTM网络中进行训练;最终得到重要用户电能质量趋势变化预测模型,并以重要电力用户的电压偏差、电压总谐波畸变率、短时间闪变等电能质量指标进行性能评估。最后,在实例分析中验证了所提出的方法的实用性和有效性,可为重要用户高品质电能质量的供应保障提供重要技术支撑。To accurately grasp the changing trend of the power quality of important users and further effectively realize a high-quality power supply to them,a method for predicting and analyzing the power quality trend of important users based on distance correlation coefficient and long short-term memory(LSTM)network is proposed in this paper.First,the raw data about the multi-dimensional feature that can characterize the power quality of important power users is stan⁃dardized,and the distance correlation coefficient is used to filter low-correlation features to achieve feature dimension reduction,so as to complete the selection of the basic sample data set of trend variation.Then,the training set samples are input into a two-layer LSTM network for training,and the prediction model for the power quality trend variation of important users is obtained eventually.Moreover,the performance evaluation is carried out based on power quality in⁃dexes such as the voltage deviation between important power users,total harmonic distortion of voltage,and short-term flicker.Finally,the practicability and effectiveness of the proposed method are verified by the analysis of an example,providing important technical support for the supply guarantee of high-quality power quality for important users.

关 键 词：距离相关系数 电能质量 长短时记忆网络 趋势变化 

分 类 号：TM933[电气工程—电力电子与电力传动]