高速铁路轨道结构性能参数动态诊断的反演理论与方法

Track structure diseases are the important factor that affects greatly the durability and service life of high-speed railway track structure, then the running safty of train. And how to diagnose dynamicly the track structure diseases is an important problem to be solved. This project, therefore, aims at proposing a new solution to find out dynamically the track structure diseases by working out its corresponding structure performance parameters while a train is running. Hopefully, the solution is to build the observation relationship between the track structure and dynamic response of the running vehicle, then to inverse the values of track structure performance parameters using the response signals of vehicle. The project involves the following researches: (1) to create the train-track-bridge system dynamics model using state-space equations, (2) to create observation transformation matrix to get the vibration behavior of track, relating to its structure performace parameters, from the response signals of vehicle, (3)to extract the response signal component causing by the changes of track structure performance parameters from the comprehensive response signals of vehicle based on blind signal algorithm, (4) to find out the change rule of track structure performance parameters and calculate out their actual values based on particle filter algorithm. Thus, this project would bulid an new inversion theroy and method of diagnosing dynamically the track structure performance parameters utilizing the difference of the field measuring signals of vehicle responses and their corresponding theory signals, which leads to an vitally important value of engineering science.

高速铁路轨道结构病害是影响轨道结构耐久性和行车安全的重要因素。轨道结构病害的动态诊断是当今急待解决的重要科学问题。本项目将轨道结构病害诊断转换为轨道结构参数的诊断，通过建立其与车辆动力响应信号之间的观测关系，研究运用车辆动力响应对轨道结构性能参数进行动态诊断的反演理论，探索一种动态诊断新方法：在信号处理及控制理论框架内，建立系统动力学模型的状态空间方程，构建观测变换矩阵，将轨道结构性能参数表达为可进行观测的车辆动力响应观测量；采用盲信号分离算法在车辆动力响应中分离出与轨道结构性能参数相关的响应分量，为轨道结构性能参数的获取与计算提供条件；采用粒子滤波算法获取结构性能参数变化特征，实现轨道结构性能诊断与分析，形成由车辆动力响应观测、处理到结构性能参数获取的诊断理论。本项目基于车辆动力响应计算实际观测量和理论响应状态量之间的差异性来反演结构特性矩阵参数，其理论研究成果具有重要的工程与科学价值。

轨道结构病害诊断技术是确保列车行车安全的关键技术，现有基于传统动力学模型响应信号分析，实现病害诊断有效性上有待突破。本项目致力于探索一种基于车辆动力响应的轨道结构病害诊断新方法，具体完成了以下关键问题的研究：. （1）基于状态空间的车线动力学建模方法，将列车-轨道-桥梁整体动力系统划分为车辆、轨道和桥梁三个子系统。利用车辆-轨道桥梁整体系统力学模型，采用了模态综合法建立了运动微分方程，通过状态变换得到基于状态空间法的车线桥耦合振动方程。（2）研究了基于Lyapunov稳定性理论的轮对振动状态及轮轨力估计的方法，设计了振动状态估计观测器及轮轨力估计观测器，实现了系统状态变量与车辆动力响应的映射关系研究。振动状态的动力学仿真计算结果与观测结果随系统运行时间增大两者趋向一致,收敛时间小于0.2S，轮轨力仿真值与估计值具有很好的吻合性，线性相关度高，理论分析结果误差小于1.00%。（3）建立了车载振动响应状态预测模型，该模型利用轨道检测车采集多参数作为分析影响车辆振动响应因素切入点，基于神经网络三种结构选择训练算法，通过训练模型的相关指标选取最优预测模型。结果表明，该方法克服了传统的多参数耦合分析方法的复杂性和不确定性，最优预测模型指标MSE可达 。（4）利用经验模态分解、希尔伯特黄变换等相关算法进行时-频分析，提取轨道状态特征信息，实现了对轨道不平顺的有害波长进行空间定位，实现了轨头焊缝、轨枕空吊病害的识别与诊断，同时应用Lamb波实现了板式无砟轨道内部脱空缺陷检测验证。. 该项目的研究有效探索了轨道结构病害的诊断模型与方法，为工程上实现轨道结构性能参数反演提供了新途径，具有重要的理论意义与实用价值。