Research On The Method Of Power Generation And Load Forecasting As Well As Scheduling Strategy Of Photovoltaic Micro Grid

The traditional energy sources are being depleting day by day with the rapid growth of energy consumption. In order to deal with this serious problem, human beings have accelerated the research and development of the renewable energy. Among them, with the advantage of easy access and green environmental protection, photovoltaic micro grid based on solar power has become a development direction for the study and utilization of the renewable energy. Because solar power is easily affected by the weather and all the objective uncontrollable factors of the external environment, photovoltaic micro grid power generation and load will show a certain degree of uncertainty, which will make difficult to realize the economical stable operation of micro grid. Therefore, it is of great significance to develop a reasonable energy scheduling strategy of the micro grid based on accurate predicted micro grid generation and load data. In this paper by using the photovoltaic micro grid as the object of study, photovoltaic power generation and load forecasting method are comprehensively and detailed studied, based on the forecasting data, photovoltaic micro grid energy scheduling strategy is deeply studied.In this paper, the basic principles of RBF neural network and grey algorithm are deeply studied, aiming at the practical problems in the photovoltaic power generation and load forecasting. The modeling methods of these two algorithms are discussed respectively. First of all, this paper introduces neural network prediction theory, discusses the PV and load prediction model construction and design idea, qualitatively analyses the connection between weather conditions and photovoltaic power generation as well as load forecasting, explores the determination and selection method of the input amount of the photovoltaic power generation and load forecasting and the method of pre-processing the input data as well as the determination of the network structure and network output, constructs the model based on RBF neural network of photovoltaic power generation and load forecasting. Secondly, the paper discusses the idea of GM(1,1) grey model, studies the pretreatment method of grey algorithm on the input data, makes a detailed analysis of the method of solving the key parameters of the model. Because of the defects of the solving process of the model parameters, it puts forward the improvement method, analyses the predictive performance of the original model and improved model. The results show that improved algorithm brings better predicted results.In order to obtain more accurate predicted results, this paper proposes a combination forecasting algorithm, studies two kinds of combination forecasting algorithms based on fixed weights and variable weights. Finally, a numerical example is used to verify and analyze the predicted results of the two algorithms, explore the advantages of the models and the respective applying occasions.Finally, this paper makes a deep analysis on the photovoltaic micro grid structure and the scheduling strategies as well as the dynamic operation cost in different operation mode, establishes the energy scheduling model of micro grid in connected mode, and proposes the optimal scheduling strategy. With the simulation, the operation cost of the micro grid with different scheduling strategies are compared and analyzed, the economics the optimal scheduling strategy brings is verified.