摘要
为了研究新型电力负荷预测方法,设计了一种基于多列深度神经网络的电力负荷预测模型。在离散数据模式下,使用差值法初步治理,通过小波变换提取其时域特征,傅里叶变换提取其频域特征。对负荷形成的时域、频域特征共8组数据进行多列深度神经网络分析,在此基础上进行一次多列神经网络分析,得到最终的叠加三角函数回归方程。通过仿真分析表明,与多项式曲线估计法和深度迭代模糊矩阵法相比,实现了预测数据质量的显著提升。模型适用于电力负荷预测任务。
A power load prediction model based on the multi-column deep neural network was designed in order to develop a new power load prediction method. In the discrete data mode, initial treatment was completed in the difference method. Time-domain features were extracted through wavelet transform and frequency domain features were extracted through Fourier transform. Time-domain and frequency-domain characteristics of the load reflected in 8 groups of data were analyzed by use of the multi-column deep neural network. On this basis, a multi-column neural network analysis was carried out to obtain a final superposition trigonometric function regression equation. Simulation analysis indicated that the quality of prediction data was significantly improved when compared with that of the polynomial curve estimation method or the deep iterative fuzzy matrix method. The proposed model could be applicable to power load prediction.
作者
童文术
王枫
周斌
黄文杰
靖海
朱小波
Tong Wenshu;Wang Feng;Zhou Bin;Huang Wenjie;Jing Hai;Zhu Xiaobo(State Grid Hubei Electric Power Co.,Ltd.,Wuhan Hubei 430077,China;Hubei Central China Electric Power Technology Development Co.,Ltd.,Wuhan Hubei 430077,China)
出处
《电气自动化》
2021年第5期34-36,68,共4页
Electrical Automation
关键词
多列神经网络
电力负荷
预测模型
仿真分析
数据特征分析
multi-column neural network
power load
prediction model
simulation analysis
data feature analysis