利用新陈代谢灰色预测、样本自适应BP 神经网络和时间序列分析分别进行风电功率实时预测和日前预测，并采用熵值取权法确定组合权重，引入自控机制，构建反馈，提出组合预测法和基于时间序列的卡尔曼滤波法。研究结果表明，组合预测模型能减少各预测点较大误差的出现，而卡尔曼滤波能大幅消减原始序列的波动影响。-Use of metabolic gray forecast, sample adaptive BP neural network and time sequence analysis respective

Haykin S. - Kalman Filtering and Neural Networks. Very good book for learning A-Haykin S. - Kalman Filtering and Neural Networks. Very good book for learning ANN