ULTRA-SHORT-TERM POWER LOAD PREDICTION OF MICRO-GRID BASED ON IMPROVED TEMPORAL CONVOLUTIONAL NETWORK
In order to improve the accuracy of ultra-short-term prediction of power load in micro-grid,feature enhancement and attention enhancement improvements are carried out on temporal convolution networks.It replaces the one-dimensional causal dilated convolution in the time convolution network with two-dimensional convolution,and uses the time pattern attention mechanism to weight the hidden layer of the time convolution network,extract the multidimensional characteristics of the load,and excavate the potential connections in the sequence.According to the improved method,the prediction model is established and a comparative experiment is carried out to verify the accuracy of the proposed method.As the results,it can effectively deal with the uncertainty of the power load,broaden the dimension of the feature vector,effectively capture the time-related characteristics in the load sequence,and improve the prediction accuracy of the power load.
electric load forecastingmicrogridsconvolutional neural networksfeature enhancementtemporal pattern attention mechanism
NETL
NSTL
万方数据
