In order to solve the problem of uneven load scheduling caused by the difficulty of short-term load forecasting in dis-tribution areas,a short-term load forecasting method for typical distribution areas based on LSTM+Attention model(short long-term memory network+Attention mechanism model)is studied.It considers the short-term load changes caused by dif-ferent change factors such as holidays,normal work and rest and temperature,determines the relationship between each load and influencing factors in the regional division unit,analyzes the load sampling data,counts the number of influencing factors of load changes according to the contribution rate of different component eigenvectors,and uses the long short-term memory net-work to fully store historical information.Based on the Attention mechanism model,the input vector,the actual state value of the hidden layer in the network and the main parameters of the regional load are determined,and the actual probability density of the normalized power load is calculated to obtain the short term load forecasting results.The experimental results show that the proposed method has fast response speed,high prediction accuracy,and ideal prediction effect and goodness of fit.
关键词
长短期记忆网络/Attention模型/短期负荷/负荷预测/配用电信息
Key words
STM network/Attention model/short term load/load forecasting/power distribution information
维普
万方数据