Research On Short-term Load Forecasting And Energy Optimal Dispatching Strategy Of User-side Microgrid

In order to realize the sustainable development of human society,the development and utilization of renewable energy has attracted more and more attention.The application of new energy power generation based on wind energy and solar energy is becoming more and more extensive,and the traditional power system is changing to a new power system with high proportion of new energy and high power electronics.The randomness and volatility of new energy power generation have a negative impact on the security,stability and power quality of the power system.Microgrid is conducive to realizing renewable energy consumption,improving power supply reliability and improving power quality.Distributed energy storage system is one of the important components of microgrid,which can play the role of shaving peaks and filling valleys and optimizing power supply quality.This thesis focuses on the short-term load forecasting of microgrids and the energy optimization scheduling strategy of distributed energy storage system,and realizes peak shaving and valley filling,and improves the economic benefits of distributed energy storage.The main research contents are as follows:（1）The main structure and working principle of the microgrid distributed energy storage system are expounded,the typical technical types and main functions of the distributed energy storage system are comprehensively analyzed,and the economic benefit mechanism of the distributed energy storage system is studied.（2）Based on the difference of load characteristics between microgrid and large grid,carry out research on load forecasting of microgrid.The fluctuation and periodicity of the power load of the microgrid are analyzed,and the load forecasting methods under different time scales are expounded.Data normalization processing method;expounds the basic process and error evaluation index of microgrid load forecasting.（3）The method of realizing microgrid load forecasting based on LSTM neural network is studied.The input variables of the neural network were determined by Pearson and MIC coefficient analysis,the number of layers and nodes of the neural network were determined by comparative analysis,and the Adam algorithm was used as a method to optimize network parameters to establish a short-term load forecasting model for microgrid;The user load data of microgrid in a certain area verifies the validity of LSTM neural network prediction modeling.The prediction results are compared with Elman and BP neural networks,which proves the advanced nature of LSTM neural network prediction.Further research adopts PCA-LSTM,LSTM-The ARIMA combination model is used for load forecasting modeling,which improves the load forecasting accuracy.（4）The energy optimization scheduling strategy of distributed energy storage system based on particle swarm algorithm is studied.The objective function,constraints and evaluation indicators of the peak-shaving and valley-filling scheduling model are constructed;for the optimal load prediction data of the LSTM-ARIMA combined model,the constant-power charging and discharging strategy and the particle swarm optimization algorithm are used to optimize the peak-shaving and valley-filling scheduling.evaluation,which verifies the effectiveness of the method in this thesis.