摘要
提出了一种用于电力系统短期负荷预报的动态神经网络模型．这种模型同时兼顾了时序法和相关法的特点，将日期特征量、气象特征量及一天的多个（动态地确定）有功负荷水平作为神经网络的输入信息，通过对输入信息动态、灵活地处理，利用有监督的学习算法对神经网络进行训练，再预测下一天相应时间点的多个有功负荷，以提高有功日负荷预测的精度和方法的适应性。
A dynamic network model is proposed for short term load forecasting.This model synthesizes the advantages of the time sequence method and the correlative method.Disposed dynamically and flexibly,the quantities representing date,meteorology and active power levels serve as samples of the model.Using the error back propagation method,the model is trained and then employed to forecast the next periodic power levels.The experiments show that the model is efficient and feasible.
出处
《湖南大学学报（自然科学版）》
EI
CAS
CSCD
1997年第1期83-86,92,共5页
Journal of Hunan University:Natural Sciences
关键词
短期
负荷预报
神经网络
经济负荷分配
电力系统
short term load forecasting,dynamic,neural network,power economic dispatch
