Short-term wind speed prediction method based on improved empirical modal decomposition and support vector machine

Abstract

The invention provides a combined short-term wind speed prediction method based on improved empirical modal decomposition CEEMDAN and a bat algorithm BA optimization support vector machine SVM. CEEMDAN is adopted to decompose an original wind speed time sequence, and a BA-SVM model is adopted to independently predict each sub-sequence obtained through decomposition; and finally, all obtained prediction results sum to obtain a wind speed prediction value. According to the invention, the original wind speed time sequence is accurately reconstructed; the modal aliasing phenomenon existing in theprior art is overcome; meanwhile, the defects of incomplete decomposition and increased calculation amount due to the fact that the reconstruction error is reduced by increasing the number of decomposition times in the prior art are remarkably overcome; parameters of a support vector machine are optimized by adopting a bat algorithm; each component is predicted by adopting a formed BA-SVM model; prediction results of the components are superposed; and the accuracy of wind speed prediction is greatly improved.

Description

[0001] field of invention [0002] The invention relates to the field of wind power generation in electric power systems, and more specifically, relates to a short-term wind speed combination prediction method based on improved empirical mode decomposition and support vector machine. Background technique [0003] With the rapid development of society, the traditional fossil energy crisis and environmental pollution are becoming more and more serious. As an important clean and sustainable power generation method, wind power has been widely used around the world. Wind power generation is to convert the kinetic energy of wind into electrical energy, and the wind in nature is intermittent and volatile, so wind power also has volatility and instability. challenge. The main factor that determines the size of wind power is the wind speed, so it is of great significance to accurately predict the wind speed. [0004] The main wind speed prediction methods in the prior art include con...

AI Technical Summary

Problems solved by technology

Among them, the continuation method simply uses the measured value of the wind speed at the current point as the predicted value of the next point. The principle is simple and easy to implement, but the error is large when predicting the time series with large randomness; the BP neural network method considers the non-linearity of the wind speed series. Linear, but its convergence speed is slow, and it is prone to problems such as difficulty in determining the network structure, over-learning, under-learning, and local minima; the Kalman filter method uses wind speed as a state variable to establish a state space, and uses the wind speed prediction value of the previous point and the current The premise of this method is to assume that the statistical characteristics of the noise are known, and this is where the difficulty lies; the support vector machine method can be better solved when dealing with small sample regression problems Non-linear problems, but parameters such as kernel function and penalty coefficient have a great influence on the prediction results of support vector machines, and when the training set is large, the operation time of support vector machines is also longer

Images