方舟,裘炜浩,季超,夏鹏飞,龚康家,周后盘
FANG Zhou,QIU Weihao,JI Chao,XIA Pengfei,GONG Kangjia,ZHOU Houpan

• 1:国网浙江省电力有限公司信息通信分公司
• 2:杭州电子科技大学

摘要(Abstract)：

线损电量产生于发电、输电、配电及用电等环节,线损电量占供电量的百分比称为线损率。针对线损率预测问题,提出了一种基于对抗生成网络与BP神经网络的低压台区线损率预测模型。选取某市低压台区线损数据作为实验数据集,经数据预处理后,通过K-Means++算法将低压台区分类,对不同类别的低压台区分别训练对抗生成网络来增加不同类别的样本数据,利用不同类别样本数据分别训练BP神经网络搭建低压台区线损预测模型。实验结果表明,与传统BP神经网络模型相比,该预测模型具有更加准确的效果,通过对抗网络增加样本数据可以有效改善低压台区线损数据量偏小的问题。
The line loss energy consumption comes from power generation, transmission, distribution and power consumption, and the percentage of the line loss power in the power supply is called the line loss rate.Aiming at the problem of line loss rate prediction, this paper proposes a low-voltage transformer area line loss prediction model based on the generative adversarial networks and BP neural network. The line loss data of a low-voltage transformer area in a city is selected as the experimental data sets. After data pre-processing, similar low-voltage transformer areas are classified by K-Means++ algorithm, and different types of low-voltage transformer areas are trained to increase different types of data by different generative adversarial networks,Finally, different types of sample data are used to train BP neural network and construct a low-voltage transformer area line loss prediction model. The experimental results show that the prediction model proposed in this paper is more accurate than the traditional BP neural network model, and it shows that increasing the sample data by generative adversarial networks can effectively improve the problem of small line loss data quantity in the low-voltage transformer area.

关键词(KeyWords)： 线损率预测;对抗生成网络;BP神经网络;K-Means++算法
line loss rate prediction;generative adversarial networks;BP neural networks;K-Means++ Algorithm

Abstract:

Keywords:

基金项目(Foundation):

作者(Author): 方舟,裘炜浩,季超,夏鹏飞,龚康家,周后盘
FANG Zhou,QIU Weihao,JI Chao,XIA Pengfei,GONG Kangjia,ZHOU Houpan

DOI: 10.19585/j.zjdl.201910008

参考文献(References)：

• [1]周金华，魏鹏飞，王超.对如何降低线损率的措施浅释[J].中小企业管理与科技（上旬刊），2010(4):236-237.
• [2]WU A,NI B 7.Line Loss Analysis and Calculation of Electric Power Systems[M].New York:John Wiley and Sons Ltd,2015.
• [3]王海燕.基于核偏最小二乘回归分析的线损率预测[J].计算机仿真，2012,29(11):323-326.
• [4]郑强.基于灰色理论的电力线损预测[J].中国水运：理论版，2007,5(1):144-145.
• [5]彭建春，李春晖，祁学红，等.基于快速独立成分分析和支持向量回归的混合馈线线损估算[J].电力系统保护与控制，2012,40(3):51-55.
• [6]黄韬，刘胜辉，谭艳娜.基于K-means聚类算法的研究[J].计算机技术与发展，2011,21(7):54-57.
• [7]ZHANG Y,HEPNER G,DENNISON P.Delineation of phenoregions in geographically diverse regions using Kmeans++clustering:a case study in the upper colorado river basin[J].Mapping Sciences&Remote Sensing,2012,49(2):163-181.
• [8]夏士雄，李文超，周勇，等.一种改进的K-means聚类算法[J].东南大学学报（英文版），2007,23(3):435-438.
• [9]王坤峰，苟超，段艳杰，等.生成式对抗网络GAN的研究进展与展望[J].自动化学报，2017,43(3):321-332.
• [10]GOODFELLOW I J,POUGET-ABADIE J,MIRZA M,et al.Generative adversarial networks[J].Advances in Neural Information Processing Systems,2014(3):2672-2680.
• [11]林懿伦，戴星原，李力，等.人工智能研究的新前线:生成式对抗网络[J].自动化学报，2018,44(5):10-27.
• [12]余思泉，韩志，唐延东，等.基于对抗生成网络的纹理合成方法[J].红外与激光工程，2018,47(2):43-48.
• [13]JING L,CHENG J H,SHI J Y,et al.Brief Introduction of Back Propagation(BP)Neural Network Algorithm and Its Improvement[J].Advance in Computer Science and Information Engineering,2012,169:553-558.
• [14]贾丽会，张修如.BP算法分析与改进[J].计算机技术与发展，2006,16(10):101-103.
• [15]明东岳，李振华，李俊，等.配网线损同期化计算与多维分析系统[J].电网与清洁能源，2019,35(4):30-36.
• [16]刘东升，代盛国，商学斌，等.基于压缩感知理论的缺失数据集下线损预测模型[J].广东电力，2019,32(2):80-86.
• [17]陈亮，吕斌斌，方勤斌，等.基于智能电表海量数据的台区运行态势分析与应用[J].电力大数据，2019,22(2):74-80.
• [18]张永凯，马欢，张玮，等.计及峰谷分时的配电线路同期月线损预测方法及应用[J].电网与清洁能源，2017,33(10):67-73.
• [19]王若丞.配电网三相不平衡对线损增加率及电压偏移的影响[J].电力工程技术，2017,36(4):131-136.
• [20]李健，庄远灿，党三磊，等.基于特征量测量的配电网实时三相线损计算[J].广东电力，2017,30(2):81-86.