基于长短期记忆神经网络的短期负荷预测

Short-term power load forecasting based on long-term memory neural network

准确的电力负荷预测是电力系统安全、稳定、经济、优质运行的前提,负荷预测的本质是通过历史数据对未来负荷情况做出预先估计。电量的快速增长和用户的多元化对负荷预测提出了更高的要求。文中提出了一种基于长短期记忆网络的负荷预测方法,利用长短期网络数据驱动和对时间序列建模强的特点,对于含非线性、不确定性的系统,提取其负荷数据中的周期特征,具有较强的自适应性。以真实数据为算例,验证了方法的有效性。

Accurate power load forecasting is the premise of power system security,stability,economy and high-quality operation.The rapid growth of electricity and the diversification of users put forward higher requirements for load forecasting.The rapid growth of power consumption and the diversification of users have put forward a higher requirement to load forecasting.Based on long-term and short-term memory network,a load forecasting method is proposed.The periodic features of the load data can be extracted in a model-free way with strong adaptability in systems with nonlinear and uncertainties with strong adaptability.Case with the real data validates the proposed approach.