Wind Farm Cluster Power Forecasting System Based On Tuple Matching

With the rapid wind power technology development and growing wind power capacity, large-scale wind power connected to the grid has brought a lot of problems and adverse effects for power system. Intermittent and random characteristics have become a major restriction factor for the development. If wind power can be effective forecasted, it would be good for scheduling department to set out day operation mode and adjust scheduling plan, reducing the situation of abandoning wind and power rationing, and to improve wind power acceptation ratio for the whole system, which would make clean energy more efficient to use. Regional wind power forecasting can not only has good practical significance, providing wind power forecasting information of the whole grid to scheduling department, but also reduce prediction error to some extent. Therefore, regional wind power prediction is a subject worthy of further study. In this background, it is necessary for wind farm cluster power prediction to study further.This paper summarized and enriched three kinds of wind power evaluation index based on its prediction research status at home and abroad and used the above index to analyze wind power temporal and spatial distribution characteristics quantitatively. Based on wind farm cluster power pattern, tuple vector warping matching was used as core algorithm of cluster power prediction system. In order to improve the speed and accurate of prediction, this paper used double-time granularity matching process and parameter correction of matching result. When constructing the practical application system of wind farm cluster power prediction, limiting wind conditions of historical wind power were considered and recognized by its time series characteristics and reconstruction methods were proposed based on wind farm power delay correlation and wind farm power bidirectional weight ratio of ARMA. Aimed at practical application of the prediction system, this paper conducted requirement analysis and determined system target, user requirement and function requirement. Then, forecasting system structure and development platform were determined, and functions and links of each module were analyzed. Based on the above analysis and designed system construction, development of practical application system was finished.Finally, the performance of data preprocessing and prediction module were analyzed and tested based on field measured data. Test results show the rationality and practicability of proposed method.