2024.1,4(1)|智能交通与数字化高速铁路沿线风速WRF集成修正预测方法段铸1,2(1.利兹大学土木工程学院,英国利兹LS29JT;2.中南大学轨道交通安全教育部重点实验室,湖南长沙410075)摘要:WRF(WeatherResearchandForecasting)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-ObjectiveEnsembleWRF),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-ObjectiveGreyWolfOptimizer)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,COmbinativeDistance-basedASsessment)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。关键词:高速铁路;风速预测;WRF;多目标优化WRFintegratedcorrectionpredictionmethodforwindspeedalonghigh‑speedrailwaysDUANZhu1,2(1.SchoolofCivilEngineering,UniversityofLeeds,Leeds,UK.LS29JT;2.KeyLaboratoryofTrafficSafetyonTrackofMinistryofEducation,CentralSouthUniversity,Changsha410075,China)Abstract:PhysicallydrivenpredictionmethodssuchasWRF(WeatherResearchandForecasting)havebeenproventobeabletoobtaineffectivewindspeedpredictionresults.ThisstudyfocusesonthesensitivityoftheWRFpredictionmethodtoinitialconditions,constructsamulti-objectiveintegratedmodelMOEnWRF(Multi-ObjectiveEnsembleWRF),andoptimizesthepredictionaccuracyofWRF.Thismethodmainlyconsistsoftwosteps.First,anevaluationindexsystemdescribingthepre‐dictionaccuracyandstabilityisconstructed,andtheMulti-ObjectiveGrayWolfOptimizationalgo‐rithm(MOGWO)isusedtosimultaneouslyoptimizethemodelaccuracyandstability,andobtainPa‐retosolutionsetofintegratedweights.Then,theCOmbinativeDistance-basedASsessmentmethod(CODAS)isusedtoevaluatetheeffectivenessoftheParetosolutionset,selecttheoptimalsolutionfromthesolutionset,andapplyittotheWRFensemblemodel.Afteractualverificationatfoursites,itisfoundthattheproposedm...