高铁沿线大风预测技术研究

风速预测是风致灾害预警的关键技术。针对高铁大风预测中延迟性和误报的问题,提出一种基于完整集合经验模态分解(CEEMDAN)和长短期记忆神经网络(LSTM)的组合预测模型对高铁沿线风速进行预测。为了减少预测模型的复杂度和提高模型预测精度,原始风速数据用CEEMDAN分解并利用样本熵(SE)理论将分解出的分量按照样本熵近似值重组成趋势、细节、随机三分量后用长短期记忆神经网络建立预测模型。以高铁沿线某段风速气象数据为例,实验结果表明,与其他预测方法相比,本方法可有效降低预测延迟性和提高预测精度,准确追踪风速的波动性和非线性非平稳的变化,性能更加优越。在高速铁路沿线大风预测中能够发挥良好的适用性,减少大风预警的误报或不报等情况的发生。

Research on prediction technology of high wind along high-speed railway

Wind speed prediction is a key technology for early warning of wind-induced disasters.Aiming at the problems of delay and false positives in high-speed rail gale warning,a combined prediction model based on complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)and long-short-term memory neural network(LSTM)was proposed to predict the wind speed along the high-speed railway.In order to reduce the complexity of the prediction model and improve the prediction accuracy of the model,the original wind speed data was decomposed with CEEMDAN and the decomposed components were reconstructed into three components of trend,detail and randomness according to the sample entropy approximation by using the sample entropy(SE)theory.Then the long short-term memory neural network was used to establish the prediction model.Taking a piece of meteorological data of wind speed along a high-speed railway as an example,the experimental results show that,compared with other prediction methods,this method can effectively reduce the prediction delay and improve the prediction accuracy,accurately track the fluctuation of wind speed and nonlinear and non-stationary changes,and the performance is more superior.It can play good applicability in the gale warning along the high-speed railway and reduce the occurrence of false alarm or non-reporting of gale warning.