| 吴涵,范元亮,林建利,李泽文,李凌斐,陈伟铭,黄兴华.基于特征融合的低压台区短期负荷预测[J].电力需求侧管理,2026,28(3):73-78 |
| 基于特征融合的低压台区短期负荷预测 |
| Short-term load prediction based on feature fusion for low-voltage distribution areas |
| 投稿时间：2025-10-28 修订日期：2025-12-17 |
| DOI：10.3969/j.issn.1009-1831.2026.03.011 |
| 中文关键词: 特征融合 主成分分析 负荷辨识 负荷预测 |
| 英文关键词: feature fusion principal component analysis load identification load forecasting |
| 基金项目:国网福建省电力有限公司科技项目(52130422003N) |
|
| 摘要点击次数: 0 |
| 全文下载次数: 0 |
| 中文摘要: |
| 负荷预测是配电台区确保电力供需平衡的重要环节,对电力系统的安全预警和稳定运行具有重要的指导意义。受多种其他因素的影响,台区负荷的直接预测模型通常具有较差的泛化能力,难以满足复杂配电台区的负荷预测需求。为了提高配电台区负荷预测方法的泛化能力,提出了一种基于多元环境特征融合的台区负荷预测方法。首先,对外在环境因素进行主成分分析,将输入变量降维并提取影响台区负荷变化的环境特征主成分;然后,基于所提的负荷辨识模型进行特征融合对台区负荷进行辨识,综合分析所得的台区负荷辨识结果和环境特征主成分;最后,基于所提的负荷预测模型进行特征融合对台区分项负荷进行预测,将分项负荷的预测结果进行线性叠加,得到配电台区负荷预测结果。选取国内某低压配电台区两年来的负荷数据作为算例,与对比方法相比,所提方法得到的负荷预测曲线更接近真实负荷数据曲线,有效提高了负荷预测的准确率。 |
| 英文摘要: |
| Load forecasting is an important aspect of ensuring power supply-demand balance in distribution station and has significant guiding significance for the safety warning and stable operation of the power system. Affected by various other factors, the direct prediction model of the load in the distribution area usually has poor generalization ability and is difficult to meet the load forecasting requirements of complex distribution substations. To improve the generalization ability of load forecasting methods in distribution areas, a load prediction method based on multi-source environmental feature fusion is proposed. Firstly, the external environmental features are analyzed by principal component analysis, and the input variables are reduced and corrected to extract the environmental feature components that affect the load changes in distribution area. Based on the proposed load identification model, the substation load is identified by feature fusion. Then, based on the comprehensive analysis of the identified substation load and environmental feature components, the proposed load forecasting model is used to predict the substation's individual load by feature fusion. Finally, the linear superposition of the predicted results of individual loads is obtained, and the load prediction result of the distribution substation is obtained. Selecting the load data of a low-voltage distribution substation in China for two years as an example, compared with other direct prediction methods, the proposed method's load prediction curve is closer to the true load data curve, effectively improving the accuracy of load forecasting. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|