【摘要】 提出一种新型的基于小世界优化的支持向量机与灰色预测变权组合风电功率预测模型。该模型发挥小世界优化算法避免陷入局部极小、快速收敛等优势,对组合权重系数进行移动样本自适应变权求解,同时,支持向量机采用实数编码小世界算法(R-SWOA)进行回归估计,构成支持向量机改进算法(RSWO-SVM)。利用江苏某风场数据对风电机组输出功率的超短期实时滚动功率预测进行研究,分别预测未来10 min、30 min和1 h的功率值。预测结果表明,无论哪个时间尺度,该文变权组合模型的预测精度均明显高于各单项、等权平均和最小方差固定权系数组合预测方法,预测误差大幅降低。更多还原

【Abstract】 A novel variable weight combination forecasting model of wind power based on support vector machine ofsmall-world optimization and grey prediction was proposed. The model conducts adaptive variable weighting solution ofmoving sample by small-world optimization algorithm which has the advantages of avoiding falling into local minima andfast convergence. A new support vector machine improved algorithm(RSWO-SVM)was proposed by using real-codingsmall world algorithm(R-SWOA)to estimate the parameters of support vector machines. The data were used to predictthe power values of next 10 minutes,30 minutes and 1 hour in a wind farm of Jiangsu. The forecast results showed that nomatter what time scaleis,the accuracy of variable weight combination prediction is significantly higher than that of anysingle forecasting method,such as the fixed-weight average and minimum variance combination method. The forecastingerror was reduced significantly.更多还原