基于多普勒激光雷达的风场预测

Application of Doppler LIDAR data in wind forecasting

采用灰色理论、BP神经网络、布朗三次指数平滑算法来对雷达风场预测进行研究。利用香港国际机场激光雷达数据对风场PPI扫描风速进行预测,得到了预测时刻机场区域整个风场的概况,让飞行员能提前了解机场区域内未来一段时间的风场环境;使用下滑道扫描数据对飞机在下滑道附近的迎头风进行预测,从而更精确地预测飞机在起飞和降落过程中将会遇到的风场波动,使飞行员能够在风切变探测系统告警之前,增加飞行员进行反应和相关准备的时间。对实验结果的统计分析表明:布朗三次指数平滑预测在延长预测时间时,预测精度随时间的增加而下降的趋势较明显;灰色预测和BP网络预测在风场变化很大的情况和延长预测时间时,仍能在一定程度上保证预测精度;灰色预测较BP网络预测效果稍好。

Using Doppler LIDAR data, three methods are implemented to forecast the wind field around Hong Kong International Airport （HKIA）. The methods are grey theory, BP neural network and triple exponential smoothing. On one hand, PPI scan data which could roughly represent the wind conditions near the arrival and departure corridors are employed to forecast the wind velocity of every point in the whole wind field. On the other hand, glide scan data are employed to obtain precise forecast of the headwind profiles,which could vividly illustrate the transient and sporadic nature of terraininduced windshear. With the forecasting results of wind field, pilots can get windshear alert in advance, so they will have more time to deal with it. Experimental results show that： （ 1 ） the performance of triple exponential smoothing degenerates with prolonging time; （2） both grey theory and BP neural network ean adapt to wind forecasting with long period, even if the wind field fluctuates fast; （ 3 ） the forecasting precision of grey theory is better than that of BP neural network.