高速铁路沿线风速WRF集成修正预测方法

WRF integrated correction prediction method for wind speed along high‑speed railways

WRF(Weather Research and Forecasting)等物理驱动预测方法被证实能够获取有效的风速预测结果。本研究面向WRF预测方法对初始条件的敏感性,构建多目标集成模型MOEnWRF(Multi-Objective Ensemble WRF),优化WRF的预测精度。该方法主要包括两个步骤,首先构建描述预测精度与稳定性的评价指标体系,采用多目标灰狼优化算法(MOGWO,Multi-Objective Grey Wolf Optimizer)实现对模型精度以及稳定性的同时优化,获得集成权重的帕累托解集。然后,采用组合距离评估法(CODAS,COmbinative Distance-based ASsessment)评估帕累托解集的有效性,从解集中选择出最优解,并应用于集成WRF模型。经过4个站点的实际验证,得到所提方法能够有效提升WRF的预测精度,并优于单目标优化集成方法。

Physically driven prediction methods such as WRF(Weather Research and Forecasting)have been proven to be able to obtain effective wind speed prediction results.This study focuses on the sensitivity of the WRF prediction method to initial conditions,constructs a multi-objective integrated model MOEnWRF(Multi-Objective Ensemble WRF),and optimizes the prediction accuracy of WRF.This method mainly consists of two steps.First,an evaluation index system describing the pre⁃diction accuracy and stability is constructed,and the Multi-Objective Gray Wolf Optimization algo⁃rithm(MOGWO)is used to simultaneously optimize the model accuracy and stability,and obtain Pa⁃reto solution set of integrated weights.Then,the COmbinative Distance-based ASsessment method(CODAS)is used to evaluate the effectiveness of the Pareto solution set,select the optimal solution from the solution set,and apply it to the WRF ensemble model.After actual verification at four sites,it is found that the proposed method can improve the prediction accuracy of WRF,and outperforms the single-objective optimization method.