摘要
现代电力负荷数据以海量形式存在,传统单机模式无法满足电力负荷在线预测效果的要求。为了改善大规模电力负荷数据的预测效果,设计了基于极限学习机的分布式电力负荷预测模型。首先提取电力负荷数据,通过混沌理论的相空间重构方法对电力负荷数据进行预处理,产生电力负荷数据预测建模样本,然后将电力负荷数据预测建模样本细分成为多个子样本,通过云计算集群系统的分布式方式并行实现子样本建模,每一个小样本通过极限学习机进行建模和预测,最后采用具体电力负荷数据进行了仿真测试实例研究,测试结果表明,本文模型加快了大规模电力负荷数据建模速度,可以足电力负荷在线预测效果,而且电力负荷预测精度要明显优于当前其它电力负荷预测模型。
Modern power load data exist in mass form, and the traditional single machine model cannot meet the demand of power load online prediction. In order to improve the prediction effect of large-scale power load data, a distributed power load forecasting model based on limit learning machine is designed. The extraction of power load data, and the load data is preprocessed by phase space reconstruction of chaotic theory, predictive modeling sample from power load data, then the data of power load prediction modeling samples are divided into multiple sub samples, through cloud computing distributed cluster parallel sub sample modeling, each a small sample the modeling and forecasting by extreme learning machine, finally through the concrete data of power load simulation test case, test results show that this model can speed up large-scale power load data modeling speed can satisfy the on-line power load prediction effect, and the accuracy of load forecasting is better than the other power load forecasting model.
作者
张丽珍
Zhang Lizhen(Electric Power Occupational Technical Institute of SEPC,Taiyuan Shanxi 030021,China)
出处
《科技通报》
2019年第2期101-105,共5页
Bulletin of Science and Technology
关键词
电力负荷
单机模式
分布式处理方式
极限学习机
云计算集群系统
power load
single machine mode
distributed processing mode
limit learning machine
cloud computing cluster system
作者简介
张丽珍(1978-),女,汉族,山西交城人,讲师,硕士,研究方向:智能电网,配电自动化。
