Research On Energy Management Strategy Of Campus Power Grid Based On Model Predictive Control Theor

In today’s industrial park grid model,the large number of renewable distributed generation and electric vehicles has increased the inaccuracy of the park grid’s dispatch.The current dispatching model only considers the use of renewable distributed generation and energy storage to meet the dispatching needs of the park grid,while the large number of electric vehicles connected to the park grid has the dual attributes of typical load and power source,when they can participate in dispatching as energy storage,it has important value for the development of the park grid.However,in the research process of the park grid,firstly,how to normalize the electric vehicles in different charging fields and form a reasonable and orderly dispatching rule,and secondly,how to deal with the reasonable dispatching among the renewable distributed generation,energy storage and loads contained in the park grid still needs further research.Based on the above problems,this paper applies the maximum charging delay index to achieve a uniform metric for electric vehicles connected to the campus grid,and applies the dynamic two-level energy management strategy of the campus grid with the model predictive control method to orderly manage the equipment of the campus grid,and builds a model predictive control model with the objective function of approaching the load forecast value of the campus grid and reducing the operating cost of the campus grid as much as possible.The main research work is as follows.For the prediction of renewable power supply power in the park grid,the Pearson correlation coefficient method is used to analyse the correlation between climate factors and renewable distributed power supply,to determine the input variables of the neural network algorithm,and based on the traditional neural network,through genetic algorithm improvement,to obtain two improved neural network algorithms including back propagation and extreme learning machine,training and prediction of these two algorithms,which are used for the application of renewable distributed generation prediction in the park grid in the later paper.This paper lays the foundation for the application of renewable distributed generation to predict power in the campus grid.For the park power grid dispatch,this paper set up a two-tier optimal control strategy for the park power grid based on model predictive control.The upper layer establishes a dynamic optimization model for scenery storage integration,and integrates the application of the park battery energy storage system and electric vehicle battery energy storage system to participate in the energy balance of the power grid.Based on the model predictive control technology,real-time dispatching commands and day-ahead dispatching plans are constantly provided with rolling corrections to resolve the source-load uncertainty of distributed clean power sources and charging sites in the park grid due to changes in the number of electric vehicles.The lower level considers both the charging needs of vehicle owners and the management needs of the upper level energy dispatch,and develops an orderly charging and discharging strategy for charging vehicles in the car parks within the park grid.In the end,the simulation example shows that the proposed strategy extends the energy storage system of the park grid,improves the accuracy of the dispatching strategy of the park grid,and increases the consumption of renewable distributed generation and the economic benefits of the park grid.