MRA与GP/MP组合预测及在陀螺仪漂移预测中的应用

研究MRA(Multi-resolution Analysis MRA)与GP/MP(Grey prediction model/Markov prediction model)组合预测及在陀螺仪漂移预测中的应用。陀螺仪漂移是各种外界环境影响下产生的综合效果,不同的外界影响产生的漂移样本频率特征是不同的,因而不适合用单一的预测模型进行预测。用小波将陀螺漂移信号进行分解,根据各子信号的频率特征选用灰色预测模型或马尔可夫预测模型分别进行预测,还原为总的预测结果。仿真试验显示,这种组合方法比起用单一的灰色马尔可夫方法可以将精度提高一倍左右,证明了这种方法的有效性。

Research on Gyro's Drift Forecasting Based on Wavelet MRA and GP/MP

A novel gyroscopic drift prediction method based on wavelet MRA was proposed. As single model is not suitable for Gyroscopic drift prediction because gyroscopic drift is influenced by different environment factors, and contains complex frequency characteristic. A wavelet decomposition was made to the original drift data and then proper method (Grey prediction model or Markov forecasting model) was chosen to make prediction of the transformed signals according to frequency characteristic. The prediction wanted could be expressed as a combination of forecasting results of the transformed signals. Simulated experiment shows that the hybrid method can increase the prediction precision comparing with Grey-Markov forecasting model.