Research on load loss prediction of distribution network outage based on hybrid neural network

In order to fully mine the relationship between temporal characteristics in load data and improve the accuracy of load forecasting, a load forecasting method based on convolutional neural networks (CNN) and gated recurrent unit (Gru) hybrid neural network is proposed. Taking date factors, climate factors and similar daily load factors as input, the sample data sets in the region are grouped by K-means clustering method; Then cnn network is used to extract the relationship between features and load in high-dimensional space, construct the high-dimensional feature vector of time series, and input the results into Gru network; Finally, the parameters of Gru network model in each group are trained and the load forecasting value is output. The results show that the proposed load forecasting method has significant advantages in forecasting accuracy and efficiency compared with long short term memory (LSTM) network model, Gru network model, cnn-lstm network model, support vector machine regression model and decision tree model.