Photovoltaic Power Prediction Considering The Influence Of Haze On Solar Radiation Intensity

The proposal of the "double carbon" goal will greatly promote the further development of the photovoltaic power generation industry.Large-scale photovoltaic power generation connection will bring a non-negligible impact to the power grid,and it is urgent to make more accurate predictions of photovoltaic power generation power to provide support for grid dispatching in advance and better and better consumption of photovoltaic power generation.In recent years,remarkable results have been achieved in the treatment of domestic atmospheric problems,but the smog problem is still an important factor affecting photovoltaic power generation.This paper carries out the research on photovoltaic power generation power prediction under haze weather,which has important practical engineering significance and important technical practical value.Firstly,various meteorological data under haze weather and their correlation with the optical thickness(AOT)of atmospheric aerosols were analyzed,and the PM2.5,PM10 and relative humidity with the greatest correlation with AOT were selected as the characteristics of clustering analysis method for clustering algorithm clustering,and the data characteristics after clustering were applied to the RBF neural network,and a multi-band AOT prediction model was trained and compared with the traditional RBF neural network and LSTM neural network.The results show that the K-means++ and RBF combined models have the highest prediction accuracy.Secondly,using the AOT data,temperature,humidity and time data output of the multi-band AOT prediction model as the input of the horizontal surface irradiance estimation model,the weight optimization is carried out based on the BAS algorithm,the horizontal surface irradiance estimation model is optimized,and the inclination surface irradiance estimation model is combined with the inclined surface irradiance estimation model,and the accuracy of the irradiance estimation model is improved by6.87% after comparison and improvement.Finally,for photovoltaic panels in haze weather of particle deposition experiment,using photovoltaic(pv)of different particles collected experimental platform sedimentary photovoltaic power under the influence of the density,the establishment and sedimentary days related to the power loss coefficient model,and improve the power loss coefficient model by means of real-time updating prediction accuracy,and photovoltaic power prediction model is established.Experimental verification shows that the average relative error of pv power prediction model considering the power loss attenuation coefficient of part.Finally,an experimental platform for photovoltaic power generation system is built,including photovoltaic power generation module,data acquisition and MPPT control module and host computer monitoring interface.The particulate matter deposition experiment of photovoltaic panels under haze weather was carried out,the photovoltaic power generation power under the influence of different particulate matter deposition density was collected,the power generation power loss coefficient model related to the number of sedimentation days was established,and the power loss coefficient model prediction accuracy was improved by real-time update,and then the photovoltaic power generation power prediction model was established.After experimental verification,the average relative error of the prediction model of photovoltaic power generation power taking into account the attenuation coefficient of particulate matter deposition power loss is 8.486%.