摘要
为有效辨识高速磁悬浮列车车轨耦合系统的关键参数,优化参数设计,亟需一种对参数灵敏度进行量化分析的方法。因此,针对参数量纲异构、时变响应全局评价困难以及传统灵敏度分析方法依赖显式函数关系的局限性,提出一种融合动态性能指标与数据驱动的改进Sobol全局灵敏度分析方法。该方法首先通过IAE、ISE、ITAE和ITSE四类时域积分指标将悬浮间隙的时变响应转化为标量化特征,消除量纲差异并整合系统动态行为的全局信息;继而利用支持向量机代理模型构建参数空间与性能指标间的隐式映射关系,突破传统方法对显式函数方程的依赖;在此基础上,结合蒙特卡罗抽样与方差分解理论,量化多参数耦合作用下各变量对系统响应方差的贡献率,实现关键参数的灵敏度排序与影响机理解析。通过“动态特征集成—代理模型逼近—方差贡献解析”的技术路径,在显著降低计算成本的同时,分析复杂参数交互效应对车轨耦合系统动态行为的影响机制,能够为工程参数优化提供理论依据。
To effectively identify the critical parameters of high-speed maglev train vehicle-rail coupling system and optimize parameter design,a method that quantifies parameter sensitivity is urgently needed.Addressing the limitations of heterogeneous parameter dimensions,difficulties in global evaluation of time-varying responses,and the reliance of traditional sensitivity analysis methods on explicit functional relationships,this study proposed an improved Sobol global sensitivity analysis method integrating dynamic performance metrics with data-driven approaches.The method first converted time-varying responses of suspension gap into scalarized features through four time-domain integral indicators(IAE,ISE,ITAE,and ITSE),eliminating dimensional differences and integrating global information of system dynamic behaviors.Subsequently,a support vector machine agent model was employed to construct implicit mapping relationships between parameter spaces and performance metrics,overcoming the dependence of traditional methods on explicit functional equations.Building on this framework,Monte Carlo sampling and variance decomposition theory were combined to quantify the contribution ratio of each variable to the system response variance under the coupling effects of multi-parameters,achieving sensitivity ranking and influence mechanism analysis of key parameters.Through the technical pathway of“dynamic feature integration-agent model approximation-variance contribution analysis”,this approach,by significantly reducing computational costs while analyzing the impact mechanisms of complex parameter interactions on the dynamic behavior of vehicle-rail coupling system,can provide a theoretical basis for engineering parameter optimization.
作者
梁仕
顾铮昊
窦峰山
龙志强
LIANG Shi;GU Zhenghao;DOU Fengshan;LONG Zhiqiang(College of Intelligence Science and Technology,National University of Defense Technology,Changsha 410073,China;National Key Laboratory of Equipment State Sensing and Smart Support,National University of Defense Technology,Changsha 410073,China)
出处
《铁道学报》
北大核心
2025年第7期136-149,共14页
Journal of the China Railway Society
基金
国家自然科学基金(52332011,52232013)。
