Load Regulation Method of Interaction Between Electric Vehicle and Power Grid Based on LSTM

Abstract

Abstract In response to the problem of disorderly charging of scaled new energy vehicles, which causes an increase in the intermittency and randomness of the overall load and then leads to the instability of the distribution network, an edge-control-based load regulation method for scaled vehicle-network interaction is proposed. Firstly, the sequence of influencing factors is constructed, and the data is processed by using weighted Marxist distance to realize the accurate extraction of electric vehicle load power data. Secondly, feature extraction of the data by using Long Short-Term Memory (LSTM) to realize fast and accurate prediction of regional load. Then, an edge-side regulation model is constructed with the objective function of minimizing network loss and minimizing peak-to-valley difference and solved by the wolf pack algorithm. Then, the edge controller is used to control the tide of each pile-end management unit to realize the effective cooperative regulation of each station-pile-net. Finally, using the edge controller physical device with an example of the distribution network in a region of Tianjin, it is verified that the method has good prediction accuracy and ensures the reasonable deployment as well as stability of the distribution network.