基于提升小波变换的风电功率预测模型

Prediction model of wind power based on lifting wavelet transform

准确的风力发电功率预测对电网的供需平衡及系统稳定运行有着重要意义。针对风电功率波动性大和随机性问题,提出一种基于提升小波变换(Lifting Wavelet Transform,LWT)的预测模型。首先,将原始风电功率数据通过提升小波算法分解成低频序列和高频序列,从而达到降低信号波动性的目的,再分别对各个子序列构建最小二乘支持向量机(Least Squares Support Vector Machine,LSSVM)模型,并考虑到LSSVM参数的选择极大程度上影响着模型的预测精度,采用改进的种群竞争算法(Improved Population Competitive Algorithm,IPCA)来优化LSSVM参数。通过数据和实际算例验证表明,采用提升小波变换进行分解明显提高了原始信号的稳定性,且相比于LSSVM和PSO-LSSVM模型,所提出的LWT-IPCA-LSSVM模型预测精度明显提高,具有理论指导意义和较好的工程应用前景。

Accurate prediction of wind power generation has a great significance for the power supply and demand balance and stable operation of power system.A prediction model based on lifting wavelet transform(LWT)is proposed to resolve the problem of large fluctuation and randomness of wind power.Firstly,the original wind power is decomposed into low and high frequency sequences by using LWT to reduce signal fluctuation.Then,the least squares support vector machines(LSSVM)model is constructed for each sequence,and considering that the choice of LSSVM parameters greatly influences the prediction accuracy of the model,an improved population competition algorithm(IPCA)is used to optimize LSSVM parameters.The simulation results show that it is effective to improve the stationarity of the original signal by using LWT,and the proposed model has higher prediction accuracy than that of LSSVM model and PSO?LSSVM model,and is of theoretical significance and good engineering application prospect.