时序数据的随机子空间模型辨识和一步预测

Stochastic Subspace Model Identification and One-step Prediction for Time Series Data

时序数据是指按先后顺序排列的一组随机数据。在系统辨识中,一组时序数据对应于随机子空间模型的输出响应。本文依照工程实际,用随机子空间辨识技术来解决时序数据的参数化建模和预测问题,提出了用改进的Positive算法进行参数识别和一步预测,来替代原有的AR参数化建模和预测技术。最后通过一组丝杆误差时序数据,对算法作了仿真和验证。

Time series data is a series of data with time sequences. In system identification, a set of time series data correspond to the output of stochastic subspace model. According to engineering practice, stochastic subspace identification technique is used to set up parameters＇ model and to forecast steps ahead for time series data. An improved positive-arithmetic is presented to identify parameters and predict one-step ahead to take the place of the traditional parameter AR（autoregression） model. Finally, the new arithmetic is simulated and validated through a set of time series data of the lathe＇s lever error.