基于CEEMDAN-WOA-SVR的高铁沿线超短期风速预测方法

Prediction Method of Ultra-Short-Term Wind Speed along High Speed Railway Based on CEEMDAN-WOA-SVR

为提升高铁沿线风速预测精度以增强铁路对强风监测预警能力,提出基于自适应噪声完备经验模态分解和鲸鱼算法优化支持向量回归(CEEMDAN-WOA-SVR)的铁路沿线风速预测方法。首先,考虑风速具有非平稳性特点和非线性趋势,基于自适应噪声完备经验模态(CEEMDAN)对风速信号进行分解,提取不同频率模态分量;其次,采用鲸鱼优化算法(WOA)优化支持向量回归(SVR)模型的惩罚因子和核参数,并构建风速预测模型;最后,以我国典型高铁沿线某测风点实测风速为例开展预测,验证风速预测方法的有效性。结果表明:所提方法对高铁沿线3 min风速预测精度较4个基准模型提升了25%,验证了它的准确性;针对5 min平均风速的预测精度提升了20%,说明它还具有较好的泛化性。该方法是对高铁沿线风速预测的有效探索,可为高铁沿线风速监测预警提供借鉴。

To improve wind speed prediction accuracy along railway lines and enhance the ability for monitoring and warning against strong winds,a hybrid model called CEEMDAN-WOA-SVR is proposed.This model is based on the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN)and uses the Whale Optimization Algorithm(WOA)to optimize Support Vector Regression(SVR).Firstly,considering the non-stationary characteristics and nonlinear trends of wind speed,CEEMDAN is applied to decompose the wind speed signal and extract modal components at different frequencies.Secondly,WOA is used to optimize the penalty factors and kernel parameters of the SVR model,and a wind speed prediction model is constructed.Finally,taking wind speed measurement points along a high-speed railway in China as an example,the prediction was carried out.The results show that the accuracy of the 3-minute wind speed prediction is improved by 25%compared to the four benchmark algorithm,thus verifying the accuracy of the method,and the prediction accuracy for the 5-minute wind speed is improved by 20%,indicating that the method has better generalization.The proposed method is an effective exploration of wind speed prediction along the high-speed railway.Therefore,the proposed model can provide reference for wind speed monitoring and warning along high-speed railways.