INTELLIGENT SYSTEM FOR FORECASTING OF THE SOLAR POWER PLANTS WORK

Abstract

The paper presents the methods and technologies of development of the computational systems for forecasting of solar power plants (SP) work in the meteorological conditions dependence. The artificial neural networks included into the analytical platform “Deductor” which has a wide range of methods for the pre-treatment of data, methods for preliminary data analysis, and data modeling are used to create forecasting systems. These systems are based on the results of the unique continuous four-year multi-parameter monitoring of SP characteristics and meteorological conditions. The following variables: voltage, current in the load circuit and power, solar radiation, external temperature, humidity, dew point, wind speed, wind direction, wind chill index, heating index, index of “temperature + humidity + wind”, index of “temperature + humidity + wind + solar radiation”, atmospheric pressure, ultraviolet index, and index of evaporation – have been registering by the monitoring system. The power density and the value of the conversion coefficient of solar energy into electrical energy are determined additionally. The paper describes two variants of the multifactor computational models for forecasting the power density and the conversion coefficient in the meteorological conditions dependence. The first variant uses the full set of recorded meteorological conditions. The second variant uses the limited set of meteorological conditions available from the forecast of the Hydro Meteorological Center of the Russian Federation. The paper gives the application examples of both variants of forecasting systems. These forecasting systems can be used not only for the direct prediction of the SP work, but also for the zoning of the Russian Federation territory from the perspectives of SP building. The importance of this approach lies in the fact that it is not a question of the generally accepted zoning of the Russian Federation territory in the solar radiation level dependence, but of zoning in terms of power density and the conversion coefficient of solar energy into electrical one. The proposed approach can be used to forecast of wind power plants work, work of thermodynamic installations using solar energy, and other devices of alternative energetic.