基于抗差自适应滤波的轨道在线检测

Rail Measurement System Based on Robust Adaptive Filter

针对测量机器人在大型导轨检测中的应用设计了一套动态跟踪检测系统。由于受各种干扰的影响,系统测量得到的测量值与目标真实状态往往不尽相同,因此提出了一种基于Singer模型的抗差自适应滤波算法。首先,采用Singer模型建立目标的运动模型;在此基础上,发展了抗差自适应滤波技术来进行在线测量数据处理;最后,进行了轨道检测的跟踪测量应用。结果证明,该系统未经滤波处理时的测量误差能够达到mm级,通过滤波能够对测量数据进行平滑处理,准确定位误差点,降低了检测误差,能够满足长轨道在线检测的准确度要求。

A kind of dynamic tracing measurement system based on geo-robot is proposed for rail measurement. Dynamic data usually suffer from various interferences and result in lower accuracy. A robust adaptive filter is proposed. Firstly, a movement model is established based on Singer model, and then a robust adaptive filter is improred to process the real measuring data. Finally, railway measurement experiment indicates that the system works efficiently, the measuring error is under mm without the RAF. With the filter, the error point can be detected accurately, and the accuracy equirement of long trail measurement is met.