摘要
针对轨道车辆齿轮箱在现有状态监测方法中面临数据样本单一与稀缺的问题,提出并构建了一种高可靠的新制式轨道车辆齿轮箱数字孪生模型,用于获取齿轮箱在多种工况下的全面、完整且可靠的数据。通过结合非线性动力学方程与动力学软件Adams的刚柔耦合仿真,建立了齿轮箱的初始孪生模型,并提出了一种基于协方差矩阵差分进化适应算法(differential evolution with covariance matrix adaptation algorithm,DE-CMAA)的高保真数字孪生建模方法。采用10维时域统计指标的敏感性分析,系统评估模型参数并筛选出关键参数。为克服现有基于余弦相似度更新方法的局限性,提出了一种结合余弦相似度和均方根误差(root mean square error,RMSE)的综合损失函数,作为DE-CMAA优化的目标函数,从而高效解决跨域模型响应差异问题。与其他2种算法相比,所提方法在相同条件下展现了更优的收敛性和更高的余弦相似度值,并在时域、频域及统计特性3个维度上展现显著的优化效果。最后,通过3种典型工况验证了该高保真孪生模型的可靠性(余弦相似度值均高于0.75),进一步证明了其在数据模拟及基于数字孪生模型的后续状态监测研究中的适用性与研究价值。
To address the limited and insufficient sample data in current state monitoring methods for railway gearboxes,this paper proposes and develops a high-reliability digital twin model for next-generation railway gearboxes.The model enables comprehensive and reliable data acquisition across various operating conditions.By integrating nonlinear dynamic equations with rigid-flexible coupling simulations in Adams,an initial twin model is built.A high-fidelity digital twin construction method based on the differential evolution with covariance matrix adaptation algorithm(DE-CMAA)is introduced.First,a 10-dimensional sensitivity analysis systematically evaluates model parameters and identifies key parameters.Then,to overcome the limitations of traditional cosine similarity-based updates,a composite loss function combining cosine similarity and root mean square error(RMSE)is proposed as the objective function for DE-CMAA optimization,effectively resolving cross-domain model response discrepancies.Compared to two other algorithms,the proposed method achieves better convergence and higher cosine similarity,demonstrating significant optimization effects across time-domain,frequency-domain,and statistical characteristics.Finally,validation under three typical operating conditions verifies the model’s reliability(Cosine similarity values exceed 0.75),showing its potential for data simulation and state monitoring in digital twin-based research.
作者
陈仁祥
冉孟宇
杨黎霞
王舜
梁栋
CHENG Renxiang;RAN Mengyu;YANG Lixia;WANG Shun;LIANG Dong(Chongqing Engineering Laboratory for Transportation Engineering Application Robot,Chongqing Jiaotong University,Chongqing 400074,China;Economics and Finance College,Chongqing University of Science&Technology,Chongqing 401331,China;Chongqing CRRC Changke Railway Vehicles Co.,Ltd.,Chongqing 401133,China)
出处
《重庆理工大学学报(自然科学)》
北大核心
2025年第8期140-147,共8页
Journal of Chongqing University of Technology:Natural Science
基金
国家自然科学基金项目(52475548)
重庆市教委科学技术研究项目(KJZD-M202200701)
重庆市自然科学基金创新发展联合基金(CSTB2023 NSCQ-LZX0127)
重庆市研究生联合培养基地项目(JDLHPYJD2024006)
重庆市研究生科研创新项目(CYS240492)。
