The Power Prediction And Control Strategy Of Photovoltaic Power Generation

With the gradual depletion of traditional fossil fuels,the global energy supply issue is still prominent.The main problems,among overall energy strategy,transformation of energy structure and dramatic energy consumption,remain to be solved.Photovoltaic power generation,as a new energy source for sustainable development of the environment,will inevitably account for a large proportion of future energy generation.However,due to the volatility and randomness of photovoltaic power,its output power is affected by various meteorological and environmental conditions as well as its own structural characteristics.The low utilization efficiency in the distribution network and excessive or insufficient output of photovoltaic power will affect the safe and reliable operation of the power grid,which also limits the integration of large-scale photovoltaic power generation systems into the power grid.Therefore,further study on the predictive model of photovoltaic power generation and the corresponding power optimization control strategy for distribution network is necessary.Firstly,this paper adopts a step-by-step predictive method to establish a prediction model for meteorological influence factors,such as the solar irradiance of photovoltaic power plants,then establish a model for generating power output characteristics of photovoltaic power plants.Since irradiance is the most important meteorological influence factor of photovoltaic power generation,its prediction accuracy has an important influence on the forecast of photovoltaic power generation.Therefore,irradiance is used as a state variable to apply the state established by the Kalman filter based on weather conditions.Spatial model,and then use Kalman filter algorithm to achieve accurate prediction of irradiance,the shortcomings of the previous irradiance prediction model is that the input space dimension is too high,resulting in a complicated structure,which brings great difficulties for learning and training.In addition,at the same time,short of meteorological factors and other information related to irradiance will also affect the prediction accuracy.Therefore,by reconstructing the input space of the irradiance prediction model,replacing the historical data sequence with statistical parameters and custom parameters reduces the dimension of the input vector,which greatly reduces the information redundancy and associated heterogeneity among the inputs.At the same time,environmental temperature,cloud thickness,and other information closely related to irradiance are introduced to improve the generalization ability and predictive accuracy of the irradiance prediction model.Then,an associated data model considering different impact factors was established and the model was applied to the comprehensive impact factor.For the input BP neural network,the weight parameters of the model are optimized by using an improved particle swarm algorithm to obtain the predicted power of the photovoltaic power generation.Finally,an active power control strategy based on the combined forecast model is proposed.This strategy uses historical data such as output power of the photovoltaic power generation system and feeder voltage to predict the output power.By analyzing the historical data of the photovoltaic power generation that causes overvoltage,the photovoltaic power generation active power that needs to be reduced is obtained.This paper proposes two kinds of power control schemes based on predictive models.The simulation model is built on Matlab and the results confirmed the feasibility and effectiveness of the two schemes under the control strategy.Scheme 3 through the energy storage device to participate in the overall control design of absorbing excess power,using the energy storage device to absorb the reduced active power.