摘要
针对传统负荷预测模型对高维非线性电力负荷的特征提取效果不理想的问题,为有效提高电力负荷短期预测精度,提出了基于模态分解-PSO-DNB深度学习的负荷预测模型。结合模态分解方法和PSO算法特征并充分融入到深度学习模型中,构造了量化深度学习模型训练效果的误差评价函数,由此建立短期负荷预测的系统模型。以某地区电网监测的电力负荷数据开展短期预测研究,通过算例效果表明,所提的预测方法可实现24 h内滚动式短期电力负荷预测,且预测误差能控制在合理范围内,相较于传统负荷预测的方法更能提升预测精度。
In order to improve the accuracy of short-term load forecasting,a load forecasting model based on mode decomposition PSO DNB deep learning is proposed.The PSO model is fully integrated into the short-term load prediction model by combining the depth learning method and the depth learning method.The short-term forecasting research is carried out based on the power load data monitored by a regional power grid.The example results show that the proposed forecasting method can realize the rolling short-term power load forecasting within 24 hours,and the forecasting error can be controlled within a reasonable range,which can improve the forecasting accuracy compared with the traditional load forecasting method.
作者
赵恩来
李向阳
王高峰
刘澎源
刘朝龙
Zhao Enlai;Li Xiangyang;Wang Gaofeng;Liu Pengyuan;Liu Chaolong(Beijing State Grid Xintong Accenture Information Technology Co.,Beijing 100053,China)
出处
《能源与环保》
2022年第5期180-186,共7页
CHINA ENERGY AND ENVIRONMENTAL PROTECTION
基金
河北电力—数字化规划设计项目(B304XA200020)。
关键词
短期负荷预测
模态分解
粒子群算法
深度学习模型
short-term load forecasting
modal decomposition
particle swarm optimization(PSO)algorithm
deep learning model
