Long-period Optimization Scheduling For Independent Microgrid Based On PV Power/Energy Forecast

With the global energy crisis and environmental pollution becoming more and more serious,the proportion of renewable energy generation represented by photovoltaic is gradually increasing in the energy structure.At present,as one of the important means to integrate multiple types of renewable energy and realize multi-energy complementarity,independent microgrid has been gradually applied to solve the power supply problems in remote areas such as islands and mountains.However,due to the influence of climate and other factors,renewable energy generation has strong randomness and uncertainty,and the long-term safe,stable and reliable operation of independent microgrid faces great challenges.To this end,this thesis focuses on photovoltaic power generation forecast and long-period optimization scheduling of independent microgrid.The main work of this thesis is summarized as follows:(1)Aiming at the problem of multi-time scale power and energy forecast for photovoltaic,analyze the influence of meteorological factors and weather types on photovoltaic output.Taking the strong correlation meteorological factors as the characteristic variables,a screening method of meteorological similarity days based on weighted grey relation analysis is proposed.Taking similarity days as training samples and BP neural network optimized by genetic algorithm as basic forecast model,the forecast models of day-ahead photovoltaic power and long-time scale photovoltaic energy are constructed respectively.(2)Aiming at the problem of large error of meteorological input source of forecast model,a modified model of total irradiance based on multi-source meteorological forecast is proposed centering on the strong correlation factor of photovoltaic output——total irradiance.The generalized weather type classification of historical data is realized based on the characteristics of power series,and the weather types forecast from public weather service are matched according to the degree of sunshine.The generalized weather type classification of numerical weather prediction is realized based on the daily mean value of the ratio of direct irradiance to total irradiance,and the systematic errors of numerical weather prediction for each generalized weather type are calculated.The tree augmented naive bayes algorithm is used to evaluate the accuracy of two meteorological forecast sources: numerical weather prediction and weather type forecast from public weather service.Combined with the above two meteorological forecast sources and the measured total irradiance,the total irradiance of numerical weather prediction is modified step by step.The modified total irradiance is used as the input of the photovoltaic power and energy forecast models to realize the day-ahead photovoltaic power forecast and long-time scale photovoltaic energy forecast.The validity of the proposed model is verified by the actual operation data and meteorological forecast data of a photovoltaic power station in Henan Province of China.(3)Aiming at the problem of long-term reliable operation of independent microgrid,a dynamic optimization method of independent microgrid considering longperiod source-charge energy matching is proposed.In the long-period energy matching evaluation and optimization stage,the energy supply and demand matching status and risk in long period are evaluated based on the photovoltaic and load energy forecast information.In addition,make the daily energy distribution scheme,from which obtain the reserve energy demand and load energy shift plan of microgrid in advance to guide the day-ahead stochastic optimization.In the day-ahead stochastic optimization stage,multi-scenario technology is applied to deal with the uncertainty of photovoltaic and load,and conditional value at risk is introduced to measure the scheduling risk caused by uncertainty.The validity of the proposed method is verified by the actual operation data of an independent microgrid in northwest China.