Study On Output Characteristics And Prediction Method Of Photovoltaic Modules Under Different Environmental Parameters

Intermittence and uncertainty of photovoltaic power generation restrict its largescale grid-connected application.Therefore,researchers have proposed to predict photovoltaic output.This will provide a reference for the operation and dispatch of power network.The analysis of photovoltaic output characteristics is the theoretical basis of photovoltaic output prediction research.In this paper,the effects of solar radiation intensity,temperature,relative humidity,wind condition and haze on photovoltaic output are analyzed in depth.This analysis lays a theoretical foundation for the study of this paper.On this basis,related research on photovoltaic output prediction methods is carried out.The main research contents are as follows:Aiming at the problem of low prediction accuracy caused by complex environmental parameters in photovoltaic output prediction methods.An improved grey BP neural network method for photovoltaic output prediction is presented in this paper.The method uses fuzzy C-means clustering algorithm to pre-classify the samples with the environmental parameters such as the highest temperature,the lowest temperature,the average temperature and the average wind speed,as well as the daily weather type as the feature of the sample classification.It chooses a kind of sample which is closest to the forecast date and obtains the forecast result through grey system.The results are used as training samples of BP neural network to establish the final model.The experiment uses the measured data of a photovoltaic power station in Jiangsu Province to verify the prediction accuracy of the method in sunny and cloudy weather conditions respectively.The prediction error analysis shows that the method reduces the prediction error to a certain extent and improves the prediction accuracy.Aiming at the complicated calculation problem caused by large data demand in traditional ultra-short-term photovoltaic output prediction method.An ultra-short-term photovoltaic output prediction method based on grey correlation analysis(GRA)and wavelet neural network(WNN)is proposed.Using GRA method,the environmental parameters such as solar radiation intensity,temperature and wind speed are taken as characteristic variables,and the samples with the greatest correlation with the predicted day are selected as training samples for WNN.Then,autocorrelation analysis is used to select the most relevant power data from the sample.The power data and the solar radiation intensity,temperature and wind speed at the corresponding time are used as input of WNN,and finally the power data at the predicted time can be obtained.The experiment uses the measured data of a photovoltaic power station in Jiangsu Province to verify the effect from the weather conditions of sunny and cloudy days.The prediction error analysis shows that the method uses less data to obtain higher prediction accuracy,which reflects its certain application value.Dust deposition in haze decreases photovoltaic output on photovoltaic modules,and fog not only weakens solar irradiation,but also intensifies dust deposition.Therefore,the influence of haze on photovoltaic output is a complicated process.In order to find out the rule of the influence of haze on photovoltaic output,a new design scheme of photovoltaic power measurement system is proposed in this paper.The system includes atomization module,high voltage xenon lamp,photovoltaic panel and MPPT controller.Among them,atomization module combined with dust injection for simulating haze,high-voltage xenon lamp is used to simulate sunlight,MPPT controller is used to detect output voltage and current and track maximum power point.The design of the system will be helpful to further study the influence of haze on photovoltaic output.In a word,compared with the corresponding traditional methods,the two photovoltaic output prediction methods studied in this paper improve the prediction accuracy and simplify the calculation process.It has certain research significance and practical value.