| 基于异构数据的电力短期负荷大数据预测方案 |
| |
| 引用本文： | 赵海波,相志军,肖林松. 基于异构数据的电力短期负荷大数据预测方案[J]. 电信科学, 2022, 38(12): 103-111. DOI: 10.11959/j.issn.1000-0801.2022292 |
| |
| 作者姓名： | 赵海波 相志军 肖林松 |
| |
| 作者单位： | 威胜信息技术股份技术有限公司,湖南长沙 410205;中国电力技术装备有限公司,北京 100052 |
| |
| 基金项目： | 国家重点研发计划项目(2020YFB0906000);国家重点研发计划项目(2020YFB0906002) |
| |
| 摘 要： | 随着多种可再生能源电力的接入，电力系统正在向更智能、更灵活、交互性更高的系统过渡。负荷预测，特别是针对单个电力客户的短期负荷预测在未来电网规划和运行中发挥着越来越重要的作用。提出了一个基于异构数据的电力短期负荷大数据预测方案，该方案收集来自智能电表和天气预报的数据，预处理后将其加载到非关系型数据库中进行存储并做进一步的异构数据处理；设计并实现了一个长短期记忆递归神经网络模型，用于确定负荷分布并预测未来24 h的住宅小区用电量；最后利用一个住宅小区的智能电表数据集对提出的短期负荷预测框架进行了测试，并使用均方根误差和平均绝对百分比误差两个指标，对比了预测模型与两种经典算法的性能，验证了所提模型的有效性。
|
| 关 键 词： | 短期负荷预测 长短期记忆网络 递归神经网络 聚类 大数据 |
A big data framework for short-term power load forecasting using heterogenous data |
| |
| Haibo ZHAO,Zhijun XIANG,Linsong XIAO. A big data framework for short-term power load forecasting using heterogenous data[J]. Telecommunications Science, 2022, 38(12): 103-111. DOI: 10.11959/j.issn.1000-0801.2022292 |
| |
| Authors: | Haibo ZHAO Zhijun XIANG Linsong XIAO |
| |
| Affiliation: | 1. Willfar Information Technology Co., Ltd., Changsha 410205, China;2. China Electronic Power Equipment and Technology Co., Ltd., Beijing 100052, China |
| |
| Abstract: | The power system is in a transition towards a more intelligent, flexible and interactive system with higher penetration of renewable energy generation, load forecasting, especially short-term load forecasting for individual electric customers plays an increasingly essential role in future grid planning and operation.A big data framework for short-term power load forcasting using heterogenous was proposed, which collected the data from smart meters and weather forecast, pre-processed and loaded it into a NoSQL database that was capable to store and further processing large volumes of heterogeneous data.Then, a long short-term memory (LSTM) recurrent neural network was designed and implemented to determine the load profiles and forecast the electricity consumption for the residential community for the next 24 hours.The proposed framework was tested with a publicly available smart meter dataset of a residential community, of which LSTM’s performance was compared with two benchmark algorithms in terms of root mean square error and mean absolute percentage error, and its validity has been verified. |
| |
| Keywords: | short-term load forecasting long short-term memory network recurrent neural network clustering big data |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《电信科学》浏览原始摘要信息 |
|
点击此处可从《电信科学》下载免费的PDF全文 |
