Super short-period wind power prediction method

Abstract

The invention provides a super short-period wind power prediction method. The method mainly comprises the following steps: S10, obtaining a wind power prediction result based on numerical weather prediction (NWP) and a wind power prediction result based on wind power farm historic / real time data, and correcting the error of the wind power prediction result based on the NWP in real time; S20, adopting an empirical distribution model, respectively building a prediction error absolute value probability distribution based on an NWP power prediction method at the time and a prediction error absolute value probability distribution based on a wind power farm historic / real time data prediction method; step 30, calculating the wind power prediction result based on the NWP at the time in a prediction time period and a weighting coefficient of the prediction result based on the wind power farm historic / real time data, and obtaining the prediction result of the time; and repeating step S20 and S30 until completing all the prediction time period, and accordingly, obtaining the prediction result of the prediction time period.

Description

technical field [0001] The invention belongs to the technical field of power system forecasting and control, and in particular relates to a comprehensive forecasting method for ultra-short-term wind power forecasting. Background technique [0002] In the context of large-scale grid-connected wind power, the volatility and randomness of wind power pose great challenges to the safe and stable operation of traditional power systems. Wind power forecasting technology is an important means to help solve this problem, among which ultra-short-term wind power forecasting technology is mainly used for real-time dispatch of wind power and online cooperation between wind power and other generating units. According to my country's power energy structure and dispatching level, the time scale of ultra-short-term wind power forecasting is generally 4 hours in the future, and the forecasting time resolution is 15 minutes. [0003] Numerical weather prediction (Numeric weather prediction, N...

AI Technical Summary

Problems solved by technology

However, since the rolling forecast frequency of NWP is generally about 6 hours, which is much higher than the rolling frequency of ultra-short-term power forecasting, wind power forecasting technology based on NWP is difficult to predict accurately in ultra-short-term forecasting; while wind power forecasting based on historical / real-time data of wind farms The statistical methods used in the technology have limited extrapolation ability, and the prediction error increases rapidly with the time scale. It is difficult to effectively improve the problem by comprehensive prediction models between different statistical methods.

At present, the accuracy of the domestic ultra-short-term wind power forecasting system is not ideal within the time scale of 1-4 hours, and the root mean square error exceeds 10%.

Images