Research On The Short-term Forecasting Techniques For Power Output Of Photovoltaic Generation System

The world’s growing energy demand raises two challenges: climate change andenergy security, and developing renewable energy has become common choice ofthe world to deal with the challenges. Solar energy is the largest available energyresource on earth so far, photovoltaic system (PV system) has advantages offlexible installation, easy maintenance, environment friendly, so it receivesattention of the world, and grid-connected PV generation system develops rapidly.Affected by solar radiation and other factors, power output of PV generation systemis fluctuant and intermittent, network operation of large-scale PV generation systemwill affect the security, stability and economic operation of power system.Accurately forecasting the short-term power output of PV generation system makesthe scheduling department of power system arrange conventional power and PVpower co-ordinately, and adjust scheduling plan reasonably, which will mitigateadverse impacts and ensure safe and stable operation of the grid when PVgeneration system accesses on the grid, which will also make full use of solarenergy resource and play the social and economic role of PV generation system.In this paper, the short-term forecasting techniques for power output of PVgeneration system have been researched. First of all, the short-term forecastingtechniques for power output of PV generation system fall into two categories: theshort-term forecasting technique based on solar radiation and model of PVgeneration system, the short-term forecasting technique based on historical data(historical meteorological data, historical solar radiation data, historical poweroutput data, etc.). Then, three parts of content have been researched in detail: theshort-term forecasting technique for solar radiation, the short-term forecastingtechnique based on solar radiation and model of PV generation system, theshort-term forecasting technique based on historical data. In the first part, theshort-term forecasting techniques for instantaneous, hourly, and daily solarradiation are studied; in the second part, the I-V characteristic model, the energyloss and conversion efficiency of MPPT, the reliability of PV generation system arestudied, and the implementation process of short-term forecasting technique basedon solar radiation and model of PV generation system is proposed; in the last part, two kinds of short-term forecasting technique based on historical data are studied,that is, the indirect forecasting method based on solar radiation, the directforecasting method based on historical power output data. Next, three typicalforecasting methods have been realized, they are, the method based on simplifiedtwo parameters DC model of solar cells, the indirect method based on improved BPneural network, and the direct method based on improved BP neural network.Finally, the programming package of short-term forecasting techniques for poweroutput of PV generation system is introduced.The short-term forecasting techniques for power output of PV generationsystem have been researched in this paper, and various forecasting methods havebeen integrated in the form of programming package finally. AS part of applicationsin energy management system of Micro-grid, the programming package issuccessfully applied to863programme and some research projects of AnhuiElectric Power Research Institute and Hefei Power Supply Company, it alsoprovides some reference and guidance to the realization of power output predictionmodule in PV’s energy management system.