Short-term Photovoltaic Generation Forecast Considering The Correlation Between Weather And Meteorological Factors

The power of photovoltaic power generation system is affected by weather,meteorological factors,geographical location and other factors.So the output power of photovoltaic power generation system is highly volatile and random.Prediction of photovoltaic power generation is a basic key technology to improve power grid control,and ensure high-ratio photovoltaic grid-connected operation.Because the output power of photovoltaic power generation system is affected by many factors,this paper puts forward a short-term photovoltaic power prediction method which is considering the correlation between weather and meteorological factors:Firstly,aiming at the problem that the data used for photovoltaic power generation prediction is affected by machine failure,power failure,etc.,i Forest algorithm is used to clear the problem data before selecting similar days.Through correlation analysis software SPSS,the Pearson correlation coefficient between photovoltaic power generation and five meteorological factors under five weather types is analyzed,and the five meteorological factors are fuzzy clustered to obtain the correlation coefficient between historical days and days to be measured.The calculation method of correlation degree between historical days and days to be measured is defined by correlation coefficient and correlation coefficient,so as to determine similar days according to the size of correlation degree.According to the photovoltaic power generation and related historical data from January 1,2019 to January 1,2020 of a photovoltaic power station in an urban area of Inner Mongolia,the advantages of the similar day selection method considering the correlation between weather and meteorological factors are analyzed by simulation.Secondly,the prediction value of short-term photovoltaic power generation requires higher accuracy and effectiveness.In this paper,ant colony algorithm is introduced to optimize the initial weights and thresholds of BP neural network to avoid falling into local optimum.Introducing adaptive learning efficiency shortens the training time and improves the training accuracy.On this basis,an ACO-BP(Ant Colony Optimization Algorithms,ACO)neural network photovoltaic power prediction model improved by adaptive learning efficiency is constructed.Compared with BP neural network,ACO-BP neural network and improved ACO-BP neural network,the second model shows that the improved ACO-BP neural network has the highest prediction accuracy.