Short-term Power Load Forecasting Based On VMD And Deep Learnin

At present,the world is facing an increasingly serious energy crisis and environmental pollution problems.For this reason,microgrid,as a small-scale power system that can connect new energy sources to the grid on a large scale,has received a lot of attention from all over the world.Microgrids convert various renewable energy sources into electricity through various distributed power sources and connect them to the grid system.Accurate microgrid load forecasting can provide a reliable basis for microgrid scheduling and improve the operational efficiency of microgrid.Therefore,the purpose of this paper is to explore a method that can accurately predict the load of microgrid.The main contents of this paper are as follows:(1)This paper first analyzes the background and research significance of this topic.The common methods and research progress of load forecasting are discussed.In addition,some related research literature is analyzed and summarized.(2)This paper gives an overview of time series decomposition methods such as Variational mode decomposition(VMD),neural networks such as Temporal Convolutional Network(TCN),and theories related to attention mechanisms,and provides the theoretical preparation for building the prediction model below.Theoretical preparation for the following prediction models.(3)In this paper,we propose the Particle Swarm Optimization(PSO)algorithm to optimize the time series decomposition method of VMD,which improves the disadvantages of VMD decomposition method that requires manual experience to assign weights,and explain its theoretical basis and algorithmic process.Subsequently,the prediction model of TCN embedded with Attention mechanism is introduced to improve the prediction accuracy and prediction efficiency of the model.And for the high stochasticity and nonlinear characteristics presented by modern power loads,a short-term power load forecasting model with PSO-optimized VMD and TCNAttention mechanism is built.Finally,the effectiveness and feasibility of the PSOoptimized VMD with TCN-Attention short-term electric load forecasting model are verified through the analysis of arithmetic examples and comparative models.(4)Based on the study of electric load,this paper takes into account the working principle and characteristics of microgrid involving power,energy storage and load,and therefore adopts an improved Attention mechanism to fully explore the coupling between "source,load and storage" in microgrid,and extracts potentially effective information to assist the prediction model in forecasting.To this end,this paper constructs a PSO-based prediction model.To this end,a load forecasting model based on PSO-optimized VMD and TCN-improved Attention is constructed.Finally,through case and comparative model analysis,it is demonstrated that the proposed microgrid load forecasting model has higher accuracy in predicting complex microgrid load characteristics compared with other comparative models.