【作者】 刘明凤； 修春波；

【Author】 LIU Mingfeng~(1,2),XIU Chunbo~(1,2) (1.Key Laboratory of Advanced Electrical Engineering and Energy Technology, Tianjin Polytechnic University,Tianjin 300387,China; 2.School of Electrical Engineering and Automation,Tianjin Polytechnic University,Tianjin 300387,China)

【机构】 天津工业大学 电工电能新技术天津市重点实验室； 天津工业大学电气工程与自动化学院；

【摘要】 为提高风速预测的准确性,采用卡尔曼滤波方法将ARMA模型和BP神经网络相结合,提出一种混合预测方法。根据时间序列分析理论,利用已知风速序列建立风速序列的自回归预测模型,并以此建立卡尔曼滤波的状态方程和测量方程。再利用BP神经网络的预测结果作为卡尔曼滤波的观测值,通过卡尔曼滤波的递推计算得到未来风速的最优估计值,从而实现风速序列的混合预测。仿真实验结果表明:混合预测方法能够有效改善风速序列的预测性能。与传统卡尔曼滤波预测结果相比,混合预测方法预测结果的延迟现象得到改善,与神经网络预测结果相比,混合预测方法在风速序列极值点的预测误差大大减小。更多还原

【Abstract】 In order to improve the prediction accuracy of wind speed series,a novel hybrid prediction method,combining ARMA model and BP neural network,based on Kalman Filter was proposed.The known wind speed series were used to establish the autoregressive model by time series analysis theory.According to the autoregressive model,the state equation and the measurement equation of Kalman filter were established.The forecast results of BP neural network were used as the observations of Kalman filter.In this way,the hybrid prediction based on Kalman filter was completed,and the forecast results of the future wind speed series were gotten by the optimal estimation of Kalman filter.Simulation results show that the hybrid prediction method can significantly improve the prediction performance of the wind speed series.The hybrid prediction method has lesser forecast delay than the conventional Kalman filter method,and smaller prediction error at the extreme points than BP neural network.更多还原