Research On Power Prediction And Active Control Of Photovoltaic Plant

At present,energy and environmental problems in the world are becoming more and more prominent.Low-carbon and sustainable development have become the consensus of global development in recent years.From the development of modern science and technology,the development and utilization of photovoltaic power generation may determine the future of human lifestyles.Photovoltaic power generation is the solar panels as energy conversion components,according to the semiconductor photovoltaic effect,the direct conversion of solar radiation into electricity.Solar energy is free,rich,clean,efficient and other advantages,precisely because of these advantages so that the rapid development of photovoltaic industry has been promoted.With the gradual increase of photovoltaic power plant capacity,photovoltaic power generation into the grid have a great impact on their power balance.Therefore,it is of great significance to adopt an effective PV power station active power control method to reduce the power fluctuation to the large power grid and enhance the schedulability.The main contents of this thesis:1)In the prediction of photovoltaic power g eneration,this paper uses the least square support vector machine(LS-SVM)photovoltaic power generation forecasting model.First,the training sample set is divided into weather types,and then the same type of the same time the historical output power value,and temperature and humidity and other meteorological information as input data.Then normalize the data and choose the kernel function.Genetic algorithm is used to select the model parameters,and the best parameters are obtained,then the least sq uares support vector machine(SVM)power prediction model is obtained.Finally,four types of weather are selected to forecast the output power of the whole point,and the prediction results are verified and analyzed.2)In the aspect of maximum power tracking,aiming at the shortcomings of fixed-step disturbance observation,this paper proposes an adaptive variable-step maximum power tracking method based on local short-circuit current starting.The MATLAB simulation model is established.The algorithm is verified that the algorithm can improve the startup speed and the sensitivity of the external environment,and adjust the disturbance step using the adaptive disturbance method to improve the power oscillation.Has a very good static and dynamic performanc e.3)In the active control strategy.Due to the solar power station active output by the environment has a lot of randomness,and distribution network load matching is dynamic.This dynamic change makes the PV power plant dispatch and power load in the distribution network have uncertainties.In order to overcome the problems of large conversion loss and low power quality of the inverter in the traditional active power control,The optimized distribution method of forecasting data shows that the PV power station has the largest output and the minimum number of inverters is the objective function.The constraints are: 1)the PV output power is greater than the minimum startup value and less than the predicted maximum power;2)Power output and grid dispatching value minimum.When the maximum total power of photovoltaic power station can be less than the dispatching value of the power grid,the maximum power tracking strategy is used to realize the full power of the power station.When the maximum total power of the photovoltaic power station is greater than the power grid dispatching value,the proposed active power distribution strategy Control each grid inverter breaking to meet the scheduling requirements.