Forecast of Power Generation for Grid-Connected PV System Based on Inclusion Degree Reasoning

Abstract

Based on the inclusion degree of reasoning, an inconsistent decision table correction algorithm and rule selection algorithm were proposed, and a forecast model of PV power generation was obtained, which was divided into rule acquisition model and forecast model. The indoor temperature, outdoor temperature and solar radiation illumination were chosen as the condition attribute, the PV array temperature and the PV power generation as the decision attribute. The decision table was established. The discretization of the decision table was completed according to the discretization method based on information entropy. Then the cubic predicted spline algorithm was used to restore the discrete predictor to continuous forecast. The forecast of the root mean square error is about 5%, and the daily forecast pass rate is 100%, which proves that the forecast model has high forecast accuracy.