Research Of Hybrid Energy Storage Control Strategy In PV Microgrid Based On Source And Load Power Prediction

In the large-scale concentrated photovoltaic power generation models,there exists the propblems such as high transmission costs,and unabsorbable energy.Hence the distributed photovoltaic power generation has become a development trend.Since the power of photovoltaic energy is random and intermittent,it will be a threat to grid stability with increasing distributed photovoltaic grid capacity,so it is urgent to propose scientific and reasonable response methods and techniques.Integrating photovoltaic,energy storage and load to form a microgrid,controlling the charging and discharging of energy storage to suppress the fluctuation of microgrid power from grid connections,and improving the "friendliness" of the photovoltaic power grid is a feasible solution.In this paper,we focus on the power demand side of the photovoltaic microgrid and select the supercapacitor and lithium iron phosphate cells to form a hybrid energy storage system to study the problems of predicting the source and load power,controlling the hybrid storage power and scheduling the hybrid storage energy.The main research contents are as follows:1.In existing photovoltaic power regression models,they do not consider the non-negativity of photovoltaic power.To address the issue,we model the photovoltaic output power model as a Poisson kernel regression problem,and approximate the Gaussian distribution with the log gamma distribution to simplify the complexity of parameter inference.Moreover,we combine the Poisson kernel regression model with sparse Bayes learning to achieve training,regression and prediction based on the sample data and ensure the non-negativity of the predicted value of the photovoltaic output power,and improve the accuracy and adaptability of the prediction method.2.To address the problem that load power cannot be accurately described by a single probability distribution,a load power model in microgrid is established based on a mixed Gaussian model.In the absence of available prior distributions,the load power prediction algorithm is studied based on the mixed Gaussian model combined with the expectation maximization algorithm,the joint maximization posterior and the maximum likelihood algorithm,respectively,and good prediction performances are obtained.In the case of available prior distributions,the accuracy of the load prediction model and the rapidity of convergence of the prediction algorithm are improved based on the mixed Gaussian model and the variational message passing algorithm.3.A parametric adaptive energy storage power control strategy based on virtual synchronous generator is proposed for the transient stability problem of microgrid.In the section,we analyze the mechanism of the virtual synchronous generator affected by the control parameters,and combine the actual demand for power regulation of energy storage system in microgrid to design a virtual synchronous generator parameter adaptive regulation method via the derivation of the equations and boundary conditions of the active-frequency characteristic shifting quadratic frequency equation.The algorithm solves the problem of fixed parameter virtual synchronous generator strategy that may lead to the energy storage output frequency and voltage overrun and improves the performance of energy storage to support microgrid transient stability.4.The hybrid energy storage control strategy in photovoltaic microgrid based on energy prediction is studied with power demand side as the application scenario.The amplitude-frequency characteristics of fluctuating power in photovoltaic microgrid are analyzed using wavelet packet algorithm,and the power allocation strategy between supercapacitor and lithium iron phosphate battery is designed.The predicted values of photovoltaic output power and load power are used to estimate the energy state of the microgrid,based on which a daily grid-connected scheduling plan is formulated,and a hybrid energy storage control strategy is proposed to track the grid-connected power curve based on energy prediction,which achieves the goal of tracking the grid-connected power scheduling curve and greatly improves the "friendliness" of grid-connected photovoltaic microgrid.The above research work takes the power demand-side photovoltaic microgrid as the application scenario,plans the optimal control problem of the hybrid energy storage system from the future time scale based on the source and load power prediction,explores a control method for the microgrid subject to grid dispatch,and provides a new idea for solving the grid-connected problem of large-scale distributed photovoltaic power generation.