An ultra-short-term wind speed prediction method

Abstract

The invention discloses an ultra-short-term wind speed prediction method, which firstly collects the original wind farm data including historical wind speed data and historical meteorological data; training samples with high similarity; further use the fast correlation filtering algorithm to select input variables and data lengths to obtain the optimal input feature set; finally, the long short-term memory neural network model is used for ultra-short-term wind speed prediction. The invention can select historical data with high similarity, and based on the fast correlation filtering algorithm, the redundancy of the input data can be reduced, and a higher ultra-short-term wind speed prediction accuracy can be achieved under the condition of less input data.

Description

technical field [0001] The invention relates to an ultra-short-term wind speed prediction method, which belongs to the technical field of wind speed prediction of wind farms. Background technique [0002] Predicting wind speed or wind power for wind farms can reduce the operating cost and spinning reserve of the power system, and provide support for the safe, economical and high-quality operation of the power system. [0003] Similarity analysis of historical data before wind speed prediction is an important means to improve the prediction accuracy. The existing similarity analysis methods are sensitive to noise and abrupt changes in wind speed. At the same time, when training the LSTM neural network, too much or too little input data may lead to large errors. The existing ultra-short-term wind speed prediction methods based on machine learning regard wind speed prediction as a static problem, however, the wind speed sequence is a dynamic time series, and the next (or multi...

AI Technical Summary

Problems solved by technology

At the same time, when training the LSTM neural network, too much or too little input data may lead to large errors.

Images