卡尔曼小波神经网络风速预测

Kalman Filter and Wavelet Neural Network Wind Speed Prediction

风速变化的随机性使得风电并网成为当今制约风电发展的瓶颈问题。如能预测风速,并提高风速预测精度,能够有助于调度部门对风电场积极进行规划和调度,减轻风电并网对电力系统产生的不利影响。用卡尔曼滤波算法建立数据滤波模型,对原始风速数据进行一级处理,去除测量误差和系统误差;再用改进的BP小波神经网络建立风速预测仿真模型;利用卡尔曼滤波后的风速数据进行风速预测,预测结果与BP神经网络风速预测方法的预测结果对比。对比结果表明该算法预测精度高,说明该算法在处理非平稳随机数据方面具有较好的应用前景。

The randomness of the wind speed makes wind power grid becomes a bottleneck which restricts the development of wind generation. If we can forecast wind speed,and improve the predictive accuracy,which can do a great deal of help to dispatch department positively on wind farm planning and scheduling,and can reduce the negative impact of the wind power grid on electric power system. Data filtering model is established by Kalman filter algorithm to give the original wind speed data primary treatment,which can eliminate the measurement error and system error of the measuring system. With improved BP wavelet neural network to construct wind speed forecasting simulation model,wind speed dates from Kalman filtering model is utilized to forecast wind speed. Compared the wind speed pre-dicting results with the simulating results getting from BP neural network prediction model,we can find that the algo-rithm given by this paper illustrates much higher accuracy,and it also shows that this prediction method has a good ap-plication prospect in the treatment of non-stationary random data.