Abstract: In order to ensure the safe operation of the power grid, the influence of meteorological and other related factors on wind power should be fully considered to predict short-term wind power. Therefore, this paper proposes a hybrid prediction method using improved Cotes Grey Model(CGM), Improved Particle Swarm Optimization(IPSO) and Least Squares Support Vector Machine(LSSVM). The CGMIPSO-LSSVM method first uses the correlation of grey model to analyze the meteorological data and other related factors at different time.Then, according to the above data, the weight of wind parameters is determined. Third, based on the above weight, the least squares support vector machine is used to estimate the wind vector. Fourth, the improved particle swarm optimization with better convergence is adopted to optimize the CGM model to obtain the final prediction result on the basis of estimated values of the wind vector. Finally, the error of the prediction result is compensated by the Fourier residual sequence. The experiment results show that the CGM-IPSO-LSSVM prediction method takes the influence of multiple factors into consideration and overcomes the problem of parameter selection optimization. It not only greatly improves prediction accuracy within the required range to provide strong basis for the scheduling of wind power integration, but also reduces the abandoned wind rate.

    Key words: short-term wind power prediction; improved cotes grey model; improved particle swarm optimization; least squares support vector machine; fusion prediction