关联维数在高速列车运行状态评估中的应用研究

基于监测数据评估高速列车空气弹簧和横向减振器等关键部件的运行状态，针对车体横向加速度振动信号，本文提出了关联维数的列车状态评估方法，分析了车体横向振动特征，对时域信号进行了频谱分析，并进一步分析了关联维数。为了计算关联维数，需先对信号进行相空间重构，然后求出重构相空间的两个关键参数；用互信息量方法[1]求出最佳延迟时间和用CAO [2]方法求出最佳嵌入维数。通过对监测数据的关联维数分析，证明了该列车在四种不同标准状态下的工况具有明显不同的关联维数特征。因此，按照关联维数的大小，就可诊断出列车可能出现的故障。研究结果表明，关联维数分析方法在设备状态监测与故障诊断中，尤其是在非线性系统的故障诊断中显示出其独特的优势，具有较为广阔的应用前景。

Applied Research of Correlation Dimension in Operating Status Evaluation of High-speed Train

On the basis of the monitoring data, the operating statuses of the key components of high speed train, such as pneumatic spring and transverse vibration absorber, etc are evaluated. In allusion to the transverse acceleration vibration signal of train body, a method for train status evaluation by using correlation dimension is put forward, the transverse vibration feature of the train body is analyzed, the spectral analysis for the time-domain signal is implemented, and the correlation dimension is further analyzed in this paper. For calculating the correlation dimension, the phase-space reconstruction for the signal needs to be implemented before, and then the two key parameters of reconstruction phase space can be determined; the optimal delay time is figured out by applying the method of mutual information, and the optimal embedded dimension is calculated by using the method of CAO. By means of ana- lyzing the correlation dimension of the monitoring data, it proves that the four working conditions in the different standard states of the train have obviously different features of the correlation dimensions; therefore the train could be diagnosed with the possibly arisen fault according to the size of the correlation dimension. The research results indicate that the analysis method with correlation dimension shows its unique superiority in condition monitoring and fault diagnosis of equipments, especially in fault diagnosis of nonlinear system, the above method has extensive application prospect.