A photovoltaic power prediction method and device
By preprocessing real-time data from photovoltaic power plants and constructing multi-dimensional features, and combining the prediction step size for power increment calculation and constraint processing, the problems of discontinuous photovoltaic power prediction results and exceeding the physical range are solved, achieving high-accuracy prediction in actual scheduling scenarios.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- LINKYOYO
- Filing Date
- 2026-02-02
- Publication Date
- 2026-06-09
AI Technical Summary
Existing photovoltaic power prediction methods are prone to producing discontinuous or physically feasible predictions when irradiance changes rapidly or when the installed capacity is close to its limit, which limits the accuracy of the predictions in actual dispatch scenarios.
By preprocessing real-time photovoltaic power data, meteorological observation data, and weather forecast data from photovoltaic power plants, power increment data is generated. A real-time feature set is generated based on multi-dimensional features, and power increment is calculated in combination with the prediction step size. Power constraints and smoothing are then performed to output the photovoltaic power prediction results.
Maintaining the continuity and physical rationality of power changes under multiple prediction steps enhances the accuracy of prediction results in actual scheduling scenarios.
Smart Images

Figure CN122178832A_ABST