Research On Forecasting Of Grid-connected Photovoltaic System Based On Multi-objective Optimization Theory

As a clean and renewable resource, solar energy has been widely used. By the end of 2013, the global photovoltaic generation capacity is 138.86 GW including new capacity 38.35 GW. However, due to the influence factors such as environment, climate, design and equipment etc. the output power of solar photovoltaic generation system is intermittent and unassured, which will affect the stability of power distribution power system.In order to achieve the reasonable use of photovoltaic power, according to the requirement of multiple objective theory, the optimization utilization of photovoltaic generation has been used as the main purpose. It sets up three subobjective considering dynamic real-time constraints of the grid-connected PV system i.e. the output power, grid power and node voltage etc. as inequality constraint conditions while regarding active power balance and the node flow equation as the equality constraint conditions. The subobjectives are the largest PV power generation based on the limited capacity, the least user cost based on different electrovalence policy and the most PV system benefits based on economic benefit and social benefit.Besides, the power forecasting model for PV power generation system based on genetic algorithm has been established in order to achieve the optimization utilization and accurate forecasting. This model consisted of data selection, genetic algorithm and data manipulation. The appropriate history data has been choosen as the initial population and use selection, crossover and mutation of genetic algorithm to complete the population genetic operations. Moreover, the optimal individual via reasonable fitness value calculation method has been selected too. At last, it obtains the forecasting results for PV output power after averaging the values of repeated measurements.In this thesis, it compared the forecasting results obtained by genetic algorithm to the actual values of operation system by using the 5.6 k W grid-connected PV power generations as research object in Beijing University of Civil Engineering and Architecture. And the root mean square error between forecasting and actual values is only in the half of NB/T 32011-2013 while the percent of pass is 100%. In conclusion, according to different PV Feed-In Tariff and electricity price policy, it puts forward the reasonable allocation and utilization of photovoltaic power generation to achieve the biggest benefits under the premise that meet the user load demand.