No abstract

Three typical models of Liu and Jordon's isotropic (extended by Klein), Hay's anisotropic model and Reindl's anisotropic model to estimate monthly-average daily total radiation on vertical surfaces with azimuth form -90° to 90° are investigated by detail. Based on the measurement of global, beam, diffused, and ground reflected radiation on the horizontal surfaces, total radiation on south-facing, east-facing and westfacing vertical surfaces for Beijing (39°48'N and 116°28'E) during the whole year (March of 2010 to February of 2011), the objective in this investigation is to perform a statistical comparison, such as, mean bias error (MBE), mean absolute percentage error (MAPE), normalized root mean square error (NRMSE) , correlation factor(CORR), to recommend one that is applicable for estimating the solar radiation on vertical surface and optimization design of building photovoltaic system for Beijing.

The rapidly growing markets for distributed and centralized grid-connected photovoltaic (PV) systems require the reliable and available information for reflecting and predicting the electricity generation of PV systems. In this work, the relationship between PV energy production and meteorological environmental factors is discussed by correlation analysis and partial correlation analysis. Meteorological data available, including the clearness index, diurnal temperature range, the global radiation on horizontal surface, and etc., are used as inputs. Then, according to factor analysis, these various interaction factors are extracted as two independent common factors. Finally, a new method based on factor analysis and multiple regression analysis has been developed for estimating the daily PV energy production. The meteorological data are collected from Wuhan Observatory, and power data from a roof grid-connected PV system located at Huazhong University of Science and Technology in Wuhan. The data of the whole year (from March in 2010 to February in 2011) has been used for model calibration and the following data of March in 2011 is used to test the predictions. The results show that there is significant positive correlation between the estimated values and the measured values; the rMBE per day is −0.14%, MAPE per day is 13.60% and rRMSE per day is 18.04%.

Solar diffuse radiation data is one of the most important parameters for assessing solar energy resource and designing grid connected photovoltaic system, due to the characteristics of ground solar radiation change significantly in recent years, previous solar beam-diffuse radiation separated model has been unable to true reflect the variation of solar radiation. A method for estimating diffuse radiation fraction Hd/H is proposed. Take Beijing as an example, according to the local measured radiation data during 10 years, with localized revising of Erbs, a new model is built. Then the relative number of sunshine hours S/S0 is introduced to establish the new regression equation with better fitting effect. Furthermore, the accuracy of model is validated using measured data consists of solar radiation and sunshine hours of 2011. Comparison of four models is performed through error analysis. It is found that the model including two variable is of high prediction accuracy. In most months, the error is lower than 10% The new model more in line with the radiation characteristics of the local actual, for photovoltaic system design has the reference significance.

No abstract