Photovoltaic Power Generation Forecasting Model with Improved Support Vector Machine Regression Based on Rough Set and Similar Day

Abstract

To diminish the effect of photovoltaic (PV) randomization on the power system, combining attribute reduction of rough set with support vector machine (SVM) regression theory, this paper applies SVM regression to directly forecast the output of the PV array, and is based on setting rough set as front-end processor and attribute reduction of historical data. According to the type of forecasting day, this paper selects multiple reasonable similar days (SD) from historical data and uses RS-SVR model to make predication. After repeated accuracy verification, the text used radial basis function as kernel function, and use parametric search and cross-validation method to determine the parameters. Finally, this paper compared average relative error of the RS-SVR forecasting model and SVR forecasting model, and verified that the RS-SVR forecasting model can effectively solve the problem of PV power output forecasting and obtain satisfactory results.