风电功率预测(wind power prediction,WPP)技术是电力系统调度与安全运行的关键性因素,为了更好地提升风电功率预测技术的精度,在集成学习的基础上提出了一种多重集成的集群短期WPP方法。所提方法包含4步:第1步,利用变分模式分解、经验...风电功率预测(wind power prediction,WPP)技术是电力系统调度与安全运行的关键性因素,为了更好地提升风电功率预测技术的精度,在集成学习的基础上提出了一种多重集成的集群短期WPP方法。所提方法包含4步:第1步,利用变分模式分解、经验模态分解和小波变换将原始风电序列分解为多个子序列;第2步,根据子序列构造多个堆叠去噪自动编码器(stacked denoising autoencoders,SDAE)进行深度学习;第3步,将第2步的结果随机划分成几个集合,利用支持向量机(support vector machine,SVM)对每个集合进行集成;第4步,将第3步的集成的结果再随机划分成几个集合,利用SVM对每个集合进行集成,重复以上步骤直至得到最终的集成预测结果。结果表明,多重集成学习得到前96 h预测结果的平均归一化均方根误差相比单次集成减少了0.0101,百分比为9.01%;相比SDAE减少了0.0151,百分比为13.54%;相比SVM减少了0.0175,百分比为14.66%。论文研究可为基于深度学习和集成学习的风电集群短期功率预测提供参考。展开更多
This paper introduces a new source separation technique exploiting the time coherence of the source signals. The proposed approach relies only on stationary second order statistics. Blind Signal Separation (BSS) metho...This paper introduces a new source separation technique exploiting the time coherence of the source signals. The proposed approach relies only on stationary second order statistics. Blind Signal Separation (BSS) method using trilinear decomposition is proposed in this paper. Simulation results reveal that our proposed algorithm has the better blind signal separation performance than joint diagonalization method. Our proposed algorithm does not require whitening processing. Moreover, our proposed algorithm works well in the underdetermined condition, where the number of sources exceeds than the number of sensors.展开更多
基金Supported by the National Natural Science Foundation of China (60801052)Aeronautical Science Foundation of China (2009ZC52036)
文摘This paper introduces a new source separation technique exploiting the time coherence of the source signals. The proposed approach relies only on stationary second order statistics. Blind Signal Separation (BSS) method using trilinear decomposition is proposed in this paper. Simulation results reveal that our proposed algorithm has the better blind signal separation performance than joint diagonalization method. Our proposed algorithm does not require whitening processing. Moreover, our proposed algorithm works well in the underdetermined condition, where the number of sources exceeds than the number of sensors.
