Research On Optimal Dispatching Of Microgrid Based On Short Term Load Forecasting

In recent years,global environmental pollution,energy crisis and other problems are imminent,the development and utilization of wind energy,solar energy and other clean energy is getting more and more attention from all over the world.The proportion of new energy generation in the power system is also increasing year by year,and microgrid is favored by countries for its high utilization of renewable energy and good controllability.Microgrid is a kind of small generation,transmission and distribution system integrating distributed power,energy storage system and users,which is the basis of future smart grid construction.However,the power of new energy generation and user load in microgrid is easily disturbed by external factors,which is very uncontrollable.To ensure the stability of microgrid power supply,load prediction is needed to get more accurate data and make reasonable power dispatching strategy.At the same time,to ensure the safe and stable operation of microgrid,the economic optimization of the system is also an important issue in the field of microgrid research.To address the above problems,this paper conducts a research based on short-term load forecasting and multi-objective optimal scheduling of microgrids.The main research contents are as follows:(1)The structure of the microgrid system studied in this paper is described,and the wind turbine,micro gas turbine,diesel generator and electric vehicle charging and discharging load models are analyzed and constructed.(2)Forecasting research based on power load data.In view of the problems of complicated correlation information between data and incomplete consideration of influencing factors in current load forecasting,analysis and research of influencing factors based on temperature,date type,weather type and historical load are conducted;in view of the problems of data loss,abnormalities and irregularities in actual data,analysis and research of data pre-processing methods are conducted;finally,Pearson correlation coefficient analysis is used to provide a basis for determining the influencing factors of data.Finally,the Pearson correlation coefficient analysis was used to provide a basis for determining the influencing factors of the data.(3)The load prediction is based on Long Short Term Memory(LSTM)neural network model.To address the problem of low prediction accuracy of this neural network model,a particle swarm based improved LSTM neural network prediction model is proposed,and the optimal values of hyperparameters in the neural network are obtained by optimizing the search with particle swarm algorithm.Finally,the neural network model is verified to have good prediction accuracy by comparison experiments.(4)A multi-objective scheduling strategy study of microgrid based on short-term load forecasting is conducted.Based on the operating characteristics of the microgrid and the load data derived from the forecasts,a hierarchical control is carried out in the EV charging and discharging dispatching part for the impact of EV charging and discharging characteristics on the microgrid,and a time-of-use tariff is used to guide users to shift EV energy consumption periods.This strategy is based on price-based demand response,establishes a dynamic time-of-use tariff model and builds an EV start-stop control strategy,and determines whether there is wind abandonment in the system by setting a threshold value in the EV controller;combines system electric load,time-of-use tariff and improved customer satisfaction,and integrates with system operation cost and carbon emission to achieve multi-objective optimal dispatch.Finally,the effectiveness of this strategy is verified by calculation examples.