Ultra Short Term Power Forecast Of Photovoltic System Based On Groud Based Cloud Image

The development and utilization of renewable energy has become a focus in the current energy field.Solar energy has the characteristics of large resource reserves,sustainable,clean,direct access and so on.Therefore,integrating photovoltaic generation system into power grid has become an important research direction to improve the energy structure of mankind and reduce carbon emissions in the process of electric energy production.At the same time,the actual irradiance of photovoltaic(PV)system has large fluctuation and uncertainty,which results in obvious stochastic fluctuation and intermittent characteristics of output power.At present,in order to deal with the stochastic volatility and intermittency of PV power generation system,the power grid mainly adopts the way of adding energy storage devices,using optional power load and PV power generation prediction.In extreme cases,the grid can also cut off its physical connection with the grid(light)to reduce its impact on the grid.PV power prediction is the most economical and practical response to the recent emergence.The previous prediction research is basically the long-term and medium-term PV power prediction.With the improvement of the regulation technology of renewable energy,it is expected to realize short-term PV prediction and ultra short term power prediction.In short-term prediction and ultra short term power prediction,cloud as the main irradiance factor must be considered.The traditional PV power prediction method can not be effectively detected.In this paper,"PV generation system considering array partial occlusion" is the research object.Firstly,this paper introduced the solar radiation in the atmospheric transmission process.In view of the direct irradiance of the ground irradiance,a clear air direct DNI prediction method based on the generalized atmospheric turbidity is presented and introduced.Considering that the cloud is an important factor in the power fluctuation of the PV power generation system,it is necessary to study the detection of the clouds.This article then introduces the content of cloud detection.Including: general information and operation environment of all sky imager and ground nephogram were introduced;it also introduces two kinds of information entropy threshold cloud detection algorithm based on sample ground nephogram,and deficiencies in the analysis of the two types of algorithms,proposes a hybrid cloud detection threshold algorithm.Then,we introduce the calculation method of the spatial distribution of PV array irradiance based on the support vector regression(SVR),which is extracted from the cloud characteristic parameters as input and clear air factors as output,which is the modeling method and implementation process of the nonlinear relation of the cloud layer to the irradiance attenuation factor from the image features of the cloud map.At the same time,through the analysis of the characteristics of the photovoltaic components,a photovoltaic array output power model considering the partial shadow occlusion is established.The overall algorithm of ultra short term power prediction of PV is formed by combining the two and cloud moving parameters.Finally,through the actual data of the PV power generation system,the prediction model of the PV system output is verified.