融合信号分解与排列熵的高铁线路风速区间预测方法

High-speed railway wind speed interval prediction method based on signal decomposition and permutation entropy

为提升高铁线路风速预测精度以保障强风下高速列车的运营安全,融合信号分解与排列熵提出高铁线路风速区间预测方法。采用信号分解技术与排列熵(PE)筛选分量,对不同复杂度分量分别利用门控循环单元分位数回归(QRGRU)和门控循环单元(GRU)构建高铁线路风速区间预测模型。对我国某高铁线路风速监测数据进行预测,结果表明:预测模型的预测精度较对比方法有显著提升,在区间预测上有良好表现,预测区间覆盖概率(PICP)、预测区间平均带宽(PINAW)及综合覆盖带宽指标(CWC)分别为94%、0.16、1.72,表明该方法能准确描述未来风速趋势,提升预测精度,对解决高铁安全运营问题具有实用价值。

In order to improve the wind speed prediction accuracy of high-speed railway lines and ensure the operation safety of high-speed trains under strong winds,a wind speed interval prediction method for high-speed railway lines is proposed by combining signal decomposition and permutation entropy.Signal decomposition technology and permutation entropy(PE)are used to screen components,and Quantile Regression with Gated Recurrent Unit(QRGRU)and Gated Recurrent Unit(GRU)are used to construct high-speed railway line wind speed interval prediction models for components of different complexity.The prediction of wind speed monitoring data of a high-speed railway line in China shows that the prediction accuracy of this prediction model is significantly improved compared with the comparison method,and it has a good performance in interval prediction,the Prediction Interval Coverage Probability(PICP),Prediction Interval Normalized Averaged Width(PINAW)and Coverage Width-based Criterion(CWC)are 94%,0.16 and 1.72 respectively.It shows that the method can accurately describe the future wind speed trend,improve the prediction accuracy,and has practical value in solving the problem of high-speed rail safety operation.