Distributed generator configuration calibration method based on TCN-BiGRU-Attention algorithm

In order to improve the efficiency of distributed generator configuration calibration, a distributed generator configuration calibration method based on attention mechanism and temporal convolutional network-bidirectional gated recurrent unit is proposed from the aspects of type identification and capacity prediction. First, the method utilizes TCN to extract features from power generation and consumption data to identify and output the types of distributed generators, followed by inputting the extracted feature matrix to capacity prediction. Then, on the basis of temporal convolutional network, a bidirectional gated loop unit and an attention mechanism are introduced for mining the generation and consumption data matrices of different distributed generators, the intrinsic correlation of different features, and highlighting the important information through the assignment. Based on the seven distributed generators data in a place in China, the experimental results show the effectiveness of the constructed prediction model; the comparison test results with traditional machine learning and deep learning prediction models show that the proposed method has higher calibration accuracy in type identification and capacity prediction.