Study On Application Of Mobile Energy Storage In Distribution Network

With the development of economy and urbanization,the government vigorously develops the distributed power,but the randomness and anti-peak load regulation characteristics of distributed power make the peak load regulation pressure of power grid increase day by day.As a solution of smoothing load,energy storage can effectively reduce load peak-valley difference.However,most of the traditional energy storage is stationary,and the flexibility of stationary energy storage in time and space is limited with the rapid change of power demand of distribution network.Therefore,mobile energy storage system emerges at the historic moment.This paper takes mobile energy storage system as the research object and studies the key technologies for its application in distribution networkFirstly,this paper establishes application scenarios of mobile energy storage on distribution network based on requirements,and analyzes functions of energy storage converters and energy storage central control units required in different scenarios.In the aspect of energy storage converter design,the grid-connected charging and discharging model of energy storage converter is built in Simulink.Based on grid model,in view of the distribution network caused by the unbalanced three-phase load current is analyzed,compared the Decouple Double Synchronous Reference Frame and Delay Signal Cancellation both positive and negative sequence current separation methods,and develop the positive and negative sequence and zero sequence current compensation strategy of control separately,the feasibility and effectiveness of the control strategy is verified by simulation.Secondly,in terms of the operation strategy of the control unit of the energy storage center,because the load forecasting can provide data support for the energy storage scheduling,the user load forecasting based on deep learning is studied.Standard RNN and LSTM are built to predict user load.Aiming at the deficiency of LSTM in long sequence,the attention mechanism is introduced,and an LSTM prediction model based on attention mechanism is proposed.The results show that the LSTM model based on attention mechanism has higher prediction accuracy than RNN and LSTM.And then,the objective function,constraint conditions and evaluation indexes of peak load shifting are determined by analyzing the model.The prediction results of LSTM model based on Attention mechanism are used as load input data of peak load shifting.The smoothing effect of constant power method,threshold method and particle swarm optimization algorithm on load after charge and discharge optimization is compared.The applicable conditions of different algorithms are analyzed.Finally,a comprehensive simulation model is established in Simulink to analyze the load improvement of distribution transformer by mobile energy storage in different application scenarios...